modelling_eda.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Load Required Libraries"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import src.util as util\n",
    "import numpy as np\n",
    "import pandas as pd\n",
    "import matplotlib.pyplot as plt\n",
    "import copy\n",
    "import hashlib\n",
    "import json\n",
    "\n",
    "from sklearn.dummy import DummyClassifier\n",
    "from sklearn.neighbors import KNeighborsClassifier\n",
    "from sklearn.tree import DecisionTreeClassifier\n",
    "from sklearn.linear_model import LogisticRegression, RidgeClassifier\n",
    "from sklearn.svm import SVC\n",
    "from sklearn.ensemble import RandomForestClassifier, AdaBoostClassifier, GradientBoostingClassifier\n",
    "from xgboost import XGBClassifier\n",
    "from sklearn.discriminant_analysis import QuadraticDiscriminantAnalysis\n",
    "from sklearn.naive_bayes import GaussianNB\n",
    "\n",
    "from sklearn.model_selection import GridSearchCV, RandomizedSearchCV\n",
    "from sklearn import tree\n",
    "from sklearn.metrics import classification_report, confusion_matrix, accuracy_score, roc_curve, auc, ConfusionMatrixDisplay\n",
    "from sklearn.utils import class_weight\n",
    "from sklearn.utils.class_weight import compute_class_weight"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Load configuration file"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "config_data = util.load_config()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Load Train and Valid set"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Load function to call dataset\n",
    "def load_train_clean(params: dict) -> pd.DataFrame:\n",
    "    x_train = util.pickle_load(params[\"train_set_clean\"][0])\n",
    "    y_train = util.pickle_load(params[\"train_set_clean\"][1])\n",
    "\n",
    "    return x_train, y_train\n",
    "\n",
    "def load_valid_clean(params:dict) -> pd.DataFrame:\n",
    "    x_valid = util.pickle_load(params[\"valid_set_clean\"][0])\n",
    "    y_valid = util.pickle_load(params[\"valid_set_clean\"][1])\n",
    "\n",
    "    return x_valid, y_valid\n",
    "\n",
    "def load_test_clean(params: dict) -> pd.DataFrame:\n",
    "    x_test = util.pickle_load(params[\"test_set_clean\"][0])\n",
    "    y_test = util.pickle_load(params[\"test_set_clean\"][1])\n",
    "\n",
    "    return x_test, y_test"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train, y_train = load_train_clean(config_data)\n",
    "x_valid, y_valid = load_valid_clean(config_data)\n",
    "x_test, y_test = load_test_clean(config_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 3. Fitting Model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.1 Baseline Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [],
   "source": [
    "def Baseline_model():\n",
    "    dummy_clf = DummyClassifier(strategy='stratified')\n",
    "\n",
    "    dummy_clf.fit(x_train[\"nonbalance\"], y_train[\"nonbalance\"])\n",
    "\n",
    "    y_pred = dummy_clf.predict(x_train[\"nonbalance\"])\n",
    "\n",
    "    report = classification_report(y_true = y_train[\"nonbalance\"],\n",
    "                                    y_pred = y_pred,\n",
    "                                    output_dict = True)\n",
    "    report = pd.DataFrame(report)\n",
    "    return report"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.5134477632112447"
      ]
     },
     "execution_count": 6,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "## Show the baseline model of raw dataset\n",
    "report = Baseline_model()\n",
    "\n",
    "report[\"macro avg\"][\"f1-score\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.2 Weight Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "def add_weight():\n",
    "    sklearn_weight = compute_class_weight(class_weight = 'balanced', \n",
    "                                            classes = np.unique(y_train[\"nonbalance\"]), \n",
    "                                            y = y_train[\"nonbalance\"])\n",
    "    sklearn_weight = dict(zip(np.unique(y_train[\"nonbalance\"]), sklearn_weight))\n",
    "\n",
    "    util.pickle_dump(sklearn_weight, config_data['model_params_path'][0])\n",
    "\n",
    "    return sklearn_weight"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{0: 0.6644518272425249, 1: 2.0202020202020203}"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sklearn_weight = add_weight()\n",
    "sklearn_weight"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.3 Baseline Model for ML model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3.1 Create Training Log"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Create Training log function\n",
    "def training_log_template() -> dict:\n",
    "    # Debug message\n",
    "    util.print_debug(\"creating training log template.\")\n",
    "\n",
    "    # Template of training log\n",
    "    logger = {\n",
    "        \"model_name\" : [],\n",
    "        \"model_uid\" : [],\n",
    "        \"training_time\" : [],\n",
    "        \"training_date\" : [],\n",
    "        \"performance\" : [],\n",
    "        \"f1_score_avg\" : [],\n",
    "         \"data_configurations\" : [],\n",
    "    }\n",
    "\n",
    "    # Debug message\n",
    "    util.print_debug(\"Training log template created.\")\n",
    "\n",
    "    # Return training log template\n",
    "    return logger"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Create training log updater\n",
    "def training_log_updater(current_log: dict, params: dict) -> list:\n",
    "    # Create copy of current log\n",
    "    current_log = copy.deepcopy(current_log)\n",
    "\n",
    "    # Path for training log file\n",
    "    log_path = params[\"training_log_path\"]\n",
    "\n",
    "    # Try to load training log file\n",
    "    try:\n",
    "        with open(log_path, \"r\") as file:\n",
    "            last_log = json.load(file)\n",
    "        file.close()\n",
    "\n",
    "    # If file not found, create a new one\n",
    "    except FileNotFoundError as fe:\n",
    "        with open(log_path, \"w\") as file:\n",
    "            file.write(\"[]\")\n",
    "        file.close()\n",
    "\n",
    "        with open(log_path, \"r\") as file:\n",
    "            last_log = json.load(file)\n",
    "        file.close()\n",
    "    \n",
    "    # Add current log to previous log\n",
    "    last_log.append(current_log)\n",
    "\n",
    "    # Save updated log\n",
    "    with open(log_path, \"w\") as file:\n",
    "        json.dump(last_log, file)\n",
    "        file.close()\n",
    "\n",
    "    # Return log\n",
    "    return last_log\n"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3.2 Create Model Object of Multiple ML Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Create model object of machine learning model\n",
    "def create_model_object(params: dict) -> list:\n",
    "    # Debug message\n",
    "    util.print_debug(\"Creating model objects.\")\n",
    "\n",
    "    # Create model objects\n",
    "    baseline_knn = KNeighborsClassifier()\n",
    "    baseline_dt = DecisionTreeClassifier()\n",
    "    baseline_logreg = LogisticRegression()\n",
    "    baseline_svm = SVC()\n",
    "    baseline_rforest = RandomForestClassifier()\n",
    "    baseline_ada = AdaBoostClassifier()\n",
    "    baseline_grad = GradientBoostingClassifier()\n",
    "    baseline_xgb = XGBClassifier()\n",
    "    baseline_qda = QuadraticDiscriminantAnalysis()\n",
    "    baseline_gnb = GaussianNB()\n",
    "\n",
    "    # Create list of model\n",
    "    list_of_model = [\n",
    "        { \"model_name\": baseline_knn.__class__.__name__, \"model_object\": baseline_knn, \"model_uid\": \"\"},\n",
    "        { \"model_name\": baseline_dt.__class__.__name__, \"model_object\": baseline_dt, \"model_uid\": \"\"},\n",
    "        { \"model_name\": baseline_logreg.__class__.__name__, \"model_object\": baseline_logreg, \"model_uid\": \"\"},\n",
    "        { \"model_name\": baseline_svm.__class__.__name__, \"model_object\": baseline_svm, \"model_uid\": \"\"},\n",
    "        { \"model_name\": baseline_rforest.__class__.__name__, \"model_object\": baseline_rforest, \"model_uid\": \"\"},\n",
    "        { \"model_name\": baseline_ada.__class__.__name__, \"model_object\": baseline_ada, \"model_uid\": \"\"},\n",
    "        { \"model_name\": baseline_grad.__class__.__name__, \"model_object\": baseline_grad, \"model_uid\": \"\"},\n",
    "        { \"model_name\": baseline_xgb.__class__.__name__, \"model_object\": baseline_xgb, \"model_uid\": \"\"},\n",
    "        { \"model_name\": baseline_qda.__class__.__name__, \"model_object\": baseline_qda, \"model_uid\": \"\"},\n",
    "        { \"model_name\": baseline_gnb.__class__.__name__, \"model_object\": baseline_gnb, \"model_uid\": \"\"},\n",
    "    ]\n",
    "\n",
    "    # Debug message\n",
    "    util.print_debug(\"Model objects created.\")\n",
    "\n",
    "    # Return the list of model\n",
    "    return list_of_model"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3.3 Create Fitting Function"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_eval(configuration_model: str, params: dict, hyperparams_model: list = None):\n",
    "\n",
    "    # Variabel to store trained models\n",
    "    list_of_trained_model = dict()\n",
    "\n",
    "    # Create log template\n",
    "    training_log = training_log_template()\n",
    "\n",
    "    # Training for every data configuration\n",
    "    for config_data in x_train:\n",
    "        # Debug message\n",
    "        util.print_debug(\"Training model based on configuration data: {}\".format(config_data))\n",
    "\n",
    "        # Create model objects\n",
    "        if hyperparams_model == None:\n",
    "            list_of_model = create_model_object(params)\n",
    "        else:\n",
    "            list_of_model = copy.deepcopy(hyperparams_model)\n",
    "\n",
    "        # Variabel to store tained model\n",
    "        trained_model = list()\n",
    "\n",
    "        # Load train data based on its configuration\n",
    "        x_train_data = x_train[config_data]\n",
    "        y_train_data = y_train[config_data]\n",
    "\n",
    "        # Train each model by current dataset configuration\n",
    "        for model in list_of_model:\n",
    "            # Debug message\n",
    "            util.print_debug(\"Training model: {}\".format(model[\"model_name\"]))\n",
    "\n",
    "            # Training\n",
    "            training_time = util.time_stamp()\n",
    "            model[\"model_object\"].fit(x_train_data, y_train_data)\n",
    "            training_time = (util.time_stamp() - training_time).total_seconds()\n",
    "\n",
    "            # Debug message\n",
    "            util.print_debug(\"Evalutaing model: {}\".format(model[\"model_name\"]))\n",
    "\n",
    "            # Evaluation\n",
    "            y_predict = model[\"model_object\"].predict(x_valid)\n",
    "            performance = classification_report(y_valid, y_predict, output_dict = True)\n",
    "\n",
    "            # Debug message\n",
    "            util.print_debug(\"Logging: {}\".format(model[\"model_name\"]))\n",
    "\n",
    "            # Create UID\n",
    "            uid = hashlib.md5(str(training_time).encode()).hexdigest()\n",
    "\n",
    "            # Assign model's UID\n",
    "            model[\"model_uid\"] = uid\n",
    "\n",
    "            # Create training log data\n",
    "            training_log[\"model_name\"].append(\"{}-{}-{}\".format(configuration_model, config_data, model[\"model_name\"]))\n",
    "            training_log[\"model_uid\"].append(uid)\n",
    "            training_log[\"training_time\"].append(training_time)\n",
    "            training_log[\"training_date\"].append(util.time_stamp())\n",
    "            training_log[\"performance\"].append(performance)\n",
    "            training_log[\"f1_score_avg\"].append(performance[\"weighted avg\"][\"f1-score\"])\n",
    "            training_log[\"data_configurations\"].append(config_data)\n",
    "\n",
    "            # Collect current trained model\n",
    "            trained_model.append(copy.deepcopy(model))\n",
    "\n",
    "            # Debug message\n",
    "            util.print_debug(\"Model {} has been trained for configuration data {}.\".format(model[\"model_name\"], config_data))\n",
    "        \n",
    "        # Collect current trained list of model\n",
    "        list_of_trained_model[config_data] = copy.deepcopy(trained_model)\n",
    "    \n",
    "    # Debug message\n",
    "    util.print_debug(\"All combination models and configuration data has been trained.\")\n",
    "    \n",
    "    # Return list trained model\n",
    "    return list_of_trained_model, training_log"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3.4 Training Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2023-01-07 22:00:17.281942 creating training log template.\n",
      "2023-01-07 22:00:17.281942 Training log template created.\n",
      "2023-01-07 22:00:17.282941 Training model based on configuration data: nonbalance\n",
      "2023-01-07 22:00:17.282941 Creating model objects.\n",
      "2023-01-07 22:00:17.282941 Model objects created.\n",
      "2023-01-07 22:00:17.283941 Training model: KNeighborsClassifier\n",
      "2023-01-07 22:00:17.294944 Evalutaing model: KNeighborsClassifier\n",
      "2023-01-07 22:00:17.361942 Logging: KNeighborsClassifier\n",
      "2023-01-07 22:00:17.364946 Model KNeighborsClassifier has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:00:17.364946 Training model: DecisionTreeClassifier\n",
      "2023-01-07 22:00:17.416940 Evalutaing model: DecisionTreeClassifier\n",
      "2023-01-07 22:00:17.440940 Logging: DecisionTreeClassifier\n",
      "2023-01-07 22:00:17.442941 Model DecisionTreeClassifier has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:00:17.443940 Training model: LogisticRegression\n",
      "2023-01-07 22:00:17.538941 Evalutaing model: LogisticRegression\n",
      "2023-01-07 22:00:17.559940 Logging: LogisticRegression\n",
      "2023-01-07 22:00:17.560940 Model LogisticRegression has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:00:17.560940 Training model: SVC\n",
      "2023-01-07 22:00:17.734121 Evalutaing model: SVC\n",
      "2023-01-07 22:00:17.800550 Logging: SVC\n",
      "2023-01-07 22:00:17.802550 Model SVC has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:00:17.803550 Training model: RandomForestClassifier\n",
      "2023-01-07 22:00:18.453037 Evalutaing model: RandomForestClassifier\n",
      "2023-01-07 22:00:18.501973 Logging: RandomForestClassifier\n",
      "2023-01-07 22:00:18.545578 Model RandomForestClassifier has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:00:18.545578 Training model: AdaBoostClassifier\n",
      "2023-01-07 22:00:18.964123 Evalutaing model: AdaBoostClassifier\n",
      "2023-01-07 22:00:19.004747 Logging: AdaBoostClassifier\n",
      "2023-01-07 22:00:19.023313 Model AdaBoostClassifier has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:00:19.023313 Training model: GradientBoostingClassifier\n",
      "2023-01-07 22:00:19.953032 Evalutaing model: GradientBoostingClassifier\n",
      "2023-01-07 22:00:19.969393 Logging: GradientBoostingClassifier\n",
      "2023-01-07 22:00:20.001737 Model GradientBoostingClassifier has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:00:20.001737 Training model: XGBClassifier\n",
      "2023-01-07 22:00:21.550714 Evalutaing model: XGBClassifier\n",
      "2023-01-07 22:00:21.584220 Logging: XGBClassifier\n",
      "2023-01-07 22:00:21.598792 Model XGBClassifier has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:00:21.598792 Training model: QuadraticDiscriminantAnalysis\n",
      "2023-01-07 22:00:21.648841 Evalutaing model: QuadraticDiscriminantAnalysis\n",
      "2023-01-07 22:00:21.669840 Logging: QuadraticDiscriminantAnalysis\n",
      "2023-01-07 22:00:21.671840 Model QuadraticDiscriminantAnalysis has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:00:21.671840 Training model: GaussianNB\n",
      "2023-01-07 22:00:21.684840 Evalutaing model: GaussianNB\n",
      "2023-01-07 22:00:21.704840 Logging: GaussianNB\n",
      "2023-01-07 22:00:21.705383 Model GaussianNB has been trained for configuration data nonbalance.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\ASUS\\anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:878: UserWarning: Variables are collinear\n",
      "  warnings.warn(\"Variables are collinear\")\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2023-01-07 22:00:21.825663 Training model based on configuration data: smote\n",
      "2023-01-07 22:00:21.825663 Creating model objects.\n",
      "2023-01-07 22:00:21.827663 Model objects created.\n",
      "2023-01-07 22:00:21.834664 Training model: KNeighborsClassifier\n",
      "2023-01-07 22:00:21.863666 Evalutaing model: KNeighborsClassifier\n",
      "2023-01-07 22:00:21.904182 Logging: KNeighborsClassifier\n",
      "2023-01-07 22:00:21.906182 Model KNeighborsClassifier has been trained for configuration data smote.\n",
      "2023-01-07 22:00:21.906182 Training model: DecisionTreeClassifier\n",
      "2023-01-07 22:00:22.000530 Evalutaing model: DecisionTreeClassifier\n",
      "2023-01-07 22:00:22.021800 Logging: DecisionTreeClassifier\n",
      "2023-01-07 22:00:22.023799 Model DecisionTreeClassifier has been trained for configuration data smote.\n",
      "2023-01-07 22:00:22.023799 Training model: LogisticRegression\n",
      "2023-01-07 22:00:22.094664 Evalutaing model: LogisticRegression\n",
      "2023-01-07 22:00:22.115754 Logging: LogisticRegression\n",
      "2023-01-07 22:00:22.116754 Model LogisticRegression has been trained for configuration data smote.\n",
      "2023-01-07 22:00:22.116754 Training model: SVC\n",
      "2023-01-07 22:00:22.602112 Evalutaing model: SVC\n",
      "2023-01-07 22:00:22.710114 Logging: SVC\n",
      "2023-01-07 22:00:22.714114 Model SVC has been trained for configuration data smote.\n",
      "2023-01-07 22:00:22.714114 Training model: RandomForestClassifier\n",
      "2023-01-07 22:00:23.927570 Evalutaing model: RandomForestClassifier\n",
      "2023-01-07 22:00:23.984573 Logging: RandomForestClassifier\n",
      "2023-01-07 22:00:24.022570 Model RandomForestClassifier has been trained for configuration data smote.\n",
      "2023-01-07 22:00:24.022570 Training model: AdaBoostClassifier\n",
      "2023-01-07 22:00:24.825006 Evalutaing model: AdaBoostClassifier\n",
      "2023-01-07 22:00:24.871004 Logging: AdaBoostClassifier\n",
      "2023-01-07 22:00:24.893274 Model AdaBoostClassifier has been trained for configuration data smote.\n",
      "2023-01-07 22:00:24.893274 Training model: GradientBoostingClassifier\n",
      "2023-01-07 22:00:27.942762 Evalutaing model: GradientBoostingClassifier\n",
      "2023-01-07 22:00:27.982760 Logging: GradientBoostingClassifier\n",
      "2023-01-07 22:00:28.060762 Model GradientBoostingClassifier has been trained for configuration data smote.\n",
      "2023-01-07 22:00:28.062764 Training model: XGBClassifier\n",
      "2023-01-07 22:00:30.879054 Evalutaing model: XGBClassifier\n",
      "2023-01-07 22:00:30.928055 Logging: XGBClassifier\n",
      "2023-01-07 22:00:30.952052 Model XGBClassifier has been trained for configuration data smote.\n",
      "2023-01-07 22:00:30.954053 Training model: QuadraticDiscriminantAnalysis\n",
      "2023-01-07 22:00:31.023052 Evalutaing model: QuadraticDiscriminantAnalysis\n",
      "2023-01-07 22:00:31.047054 Logging: QuadraticDiscriminantAnalysis\n",
      "2023-01-07 22:00:31.049051 Model QuadraticDiscriminantAnalysis has been trained for configuration data smote.\n",
      "2023-01-07 22:00:31.050052 Training model: GaussianNB\n",
      "2023-01-07 22:00:31.076055 Evalutaing model: GaussianNB\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\ASUS\\anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:878: UserWarning: Variables are collinear\n",
      "  warnings.warn(\"Variables are collinear\")\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2023-01-07 22:00:31.109055 Logging: GaussianNB\n",
      "2023-01-07 22:00:31.110054 Model GaussianNB has been trained for configuration data smote.\n",
      "2023-01-07 22:00:31.383053 Training model based on configuration data: oversampling\n",
      "2023-01-07 22:00:31.383053 Creating model objects.\n",
      "2023-01-07 22:00:31.383053 Model objects created.\n",
      "2023-01-07 22:00:31.393051 Training model: KNeighborsClassifier\n",
      "2023-01-07 22:00:31.422052 Evalutaing model: KNeighborsClassifier\n",
      "2023-01-07 22:00:31.463054 Logging: KNeighborsClassifier\n",
      "2023-01-07 22:00:31.466053 Model KNeighborsClassifier has been trained for configuration data oversampling.\n",
      "2023-01-07 22:00:31.466053 Training model: DecisionTreeClassifier\n",
      "2023-01-07 22:00:31.540769 Evalutaing model: DecisionTreeClassifier\n",
      "2023-01-07 22:00:31.561648 Logging: DecisionTreeClassifier\n",
      "2023-01-07 22:00:31.563646 Model DecisionTreeClassifier has been trained for configuration data oversampling.\n",
      "2023-01-07 22:00:31.563646 Training model: LogisticRegression\n",
      "2023-01-07 22:00:31.653586 Evalutaing model: LogisticRegression\n",
      "2023-01-07 22:00:31.678080 Logging: LogisticRegression\n",
      "2023-01-07 22:00:31.680995 Model LogisticRegression has been trained for configuration data oversampling.\n",
      "2023-01-07 22:00:31.681342 Training model: SVC\n",
      "2023-01-07 22:00:32.151258 Evalutaing model: SVC\n",
      "2023-01-07 22:00:32.307696 Logging: SVC\n",
      "2023-01-07 22:00:32.311697 Model SVC has been trained for configuration data oversampling.\n",
      "2023-01-07 22:00:32.312695 Training model: RandomForestClassifier\n",
      "2023-01-07 22:00:34.240169 Evalutaing model: RandomForestClassifier\n",
      "2023-01-07 22:00:34.476172 Logging: RandomForestClassifier\n",
      "2023-01-07 22:00:34.703171 Model RandomForestClassifier has been trained for configuration data oversampling.\n",
      "2023-01-07 22:00:34.703171 Training model: AdaBoostClassifier\n",
      "2023-01-07 22:00:36.785279 Evalutaing model: AdaBoostClassifier\n",
      "2023-01-07 22:00:36.972277 Logging: AdaBoostClassifier\n",
      "2023-01-07 22:00:37.008275 Model AdaBoostClassifier has been trained for configuration data oversampling.\n",
      "2023-01-07 22:00:37.009275 Training model: GradientBoostingClassifier\n",
      "2023-01-07 22:00:38.847631 Evalutaing model: GradientBoostingClassifier\n",
      "2023-01-07 22:00:38.864576 Logging: GradientBoostingClassifier\n",
      "2023-01-07 22:00:38.901572 Model GradientBoostingClassifier has been trained for configuration data oversampling.\n",
      "2023-01-07 22:00:38.903572 Training model: XGBClassifier\n",
      "2023-01-07 22:00:42.590566 Evalutaing model: XGBClassifier\n",
      "2023-01-07 22:00:42.628567 Logging: XGBClassifier\n",
      "2023-01-07 22:00:42.644567 Model XGBClassifier has been trained for configuration data oversampling.\n",
      "2023-01-07 22:00:42.644567 Training model: QuadraticDiscriminantAnalysis\n",
      "2023-01-07 22:00:42.696569 Evalutaing model: QuadraticDiscriminantAnalysis\n",
      "2023-01-07 22:00:42.717566 Logging: QuadraticDiscriminantAnalysis\n",
      "2023-01-07 22:00:42.718566 Model QuadraticDiscriminantAnalysis has been trained for configuration data oversampling.\n",
      "2023-01-07 22:00:42.718566 Training model: GaussianNB\n",
      "2023-01-07 22:00:42.735567 Evalutaing model: GaussianNB\n",
      "2023-01-07 22:00:42.758510 Logging: GaussianNB\n",
      "2023-01-07 22:00:42.759509 Model GaussianNB has been trained for configuration data oversampling.\n"
     ]
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "C:\\Users\\ASUS\\anaconda3\\lib\\site-packages\\sklearn\\discriminant_analysis.py:878: UserWarning: Variables are collinear\n",
      "  warnings.warn(\"Variables are collinear\")\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2023-01-07 22:00:42.886695 All combination models and configuration data has been trained.\n"
     ]
    }
   ],
   "source": [
    "list_of_trained_model, training_log = train_eval(\"\", config_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3.5 Create function to save production model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_production_model(list_of_model, training_log, params):\n",
    "    # Create copy list of model\n",
    "    list_of_model = copy.deepcopy(list_of_model)\n",
    "    \n",
    "    # Debug message\n",
    "    util.print_debug(\"Choosing model by metrics score.\")\n",
    "\n",
    "    # Create required predefined variabel\n",
    "    curr_production_model = None\n",
    "    prev_production_model = None\n",
    "    production_model_log = None\n",
    "\n",
    "    # Debug message\n",
    "    util.print_debug(\"Converting training log type of data from dict to dataframe.\")\n",
    "\n",
    "    # Convert dictionary to pandas for easy operation\n",
    "    training_log = pd.DataFrame(copy.deepcopy(training_log))\n",
    "\n",
    "    # Debug message\n",
    "    util.print_debug(\"Trying to load previous production model.\")\n",
    "\n",
    "    # Check if there is a previous production model\n",
    "    try:\n",
    "        prev_production_model = util.pickle_load(params[\"production_model_path\"])\n",
    "        util.print_debug(\"Previous production model loaded.\")\n",
    "\n",
    "    except FileNotFoundError as fe:\n",
    "        util.print_debug(\"No previous production model detected, choosing best model only from current trained model.\")\n",
    "\n",
    "    # If previous production model detected:\n",
    "    if prev_production_model != None:\n",
    "        # Debug message\n",
    "        util.print_debug(\"Loading validation data.\")\n",
    "        x_valid, y_valid\n",
    "        \n",
    "        # Debug message\n",
    "        util.print_debug(\"Checking compatibilty previous production model's input with current train data's features.\")\n",
    "\n",
    "        # Check list features of previous production model and current dataset\n",
    "        production_model_features = set(prev_production_model[\"model_data\"][\"model_object\"].feature_names_in_)\n",
    "        current_dataset_features = set(x_valid.columns)\n",
    "        number_of_different_features = len((production_model_features - current_dataset_features) | (current_dataset_features - production_model_features))\n",
    "\n",
    "        # If feature matched:\n",
    "        if number_of_different_features == 0:\n",
    "            # Debug message\n",
    "            util.print_debug(\"Features compatible.\")\n",
    "\n",
    "            # Debug message\n",
    "            util.print_debug(\"Reassesing previous model performance using current validation data.\")\n",
    "\n",
    "            # Re-predict previous production model to provide valid metrics compared to other current models\n",
    "            y_pred = prev_production_model[\"model_data\"][\"model_object\"].predict(x_valid)\n",
    "\n",
    "            # Re-asses prediction result\n",
    "            eval_res = classification_report(y_valid, y_pred, output_dict = True)\n",
    "\n",
    "            # Debug message\n",
    "            util.print_debug(\"Assessing complete.\")\n",
    "\n",
    "            # Debug message\n",
    "            util.print_debug(\"Storing new metrics data to previous model structure.\")\n",
    "\n",
    "            # Update their performance log\n",
    "            prev_production_model[\"model_log\"][\"performance\"] = eval_res\n",
    "            prev_production_model[\"model_log\"][\"f1_score_avg\"] = eval_res[\"weighted avg\"][\"recall\"]\n",
    "\n",
    "            # Debug message\n",
    "            util.print_debug(\"Adding previous model data to current training log and list of model\")\n",
    "\n",
    "            # Added previous production model log to current logs to compere who has the greatest f1 score\n",
    "            training_log = pd.concat([training_log, pd.DataFrame([prev_production_model[\"model_log\"]])])\n",
    "\n",
    "            # Added previous production model to current list of models to choose from if it has the greatest f1 score\n",
    "            list_of_model[\"prev_production_model\"] = [copy.deepcopy(prev_production_model[\"model_data\"])]\n",
    "        else:\n",
    "            # To indicate that we are not using previous production model\n",
    "            prev_production_model = None\n",
    "\n",
    "            # Debug message\n",
    "            util.print_debug(\"Different features between production model with current dataset is detected, ignoring production dataset.\")\n",
    "\n",
    "    # Debug message\n",
    "    util.print_debug(\"Sorting training log by f1 macro avg and training time.\")\n",
    "\n",
    "    # Sort training log by f1 score macro avg and trining time\n",
    "    best_model_log = training_log.sort_values([\"f1_score_avg\", \"training_time\"], ascending = [False, True]).iloc[0]\n",
    "    \n",
    "    # Debug message\n",
    "    util.print_debug(\"Searching model data based on sorted training log.\")\n",
    "\n",
    "    # Get model object with greatest f1 score macro avg by using UID\n",
    "    for configuration_data in list_of_model:\n",
    "        for model_data in list_of_model[configuration_data]:\n",
    "            if model_data[\"model_uid\"] == best_model_log[\"model_uid\"]:\n",
    "                curr_production_model = dict()\n",
    "                curr_production_model[\"model_data\"] = copy.deepcopy(model_data)\n",
    "                curr_production_model[\"model_log\"] = copy.deepcopy(best_model_log.to_dict())\n",
    "                curr_production_model[\"model_log\"][\"model_name\"] = \"Production-{}\".format(curr_production_model[\"model_data\"][\"model_name\"])\n",
    "                curr_production_model[\"model_log\"][\"training_date\"] = str(curr_production_model[\"model_log\"][\"training_date\"])\n",
    "                production_model_log = training_log_updater(curr_production_model[\"model_log\"], params)\n",
    "                break\n",
    "    \n",
    "    # In case UID not found\n",
    "    if curr_production_model == None:\n",
    "        raise RuntimeError(\"The best model not found in your list of model.\")\n",
    "    \n",
    "    # Debug message\n",
    "    util.print_debug(\"Model chosen.\")\n",
    "\n",
    "    # Dump chosen production model\n",
    "    util.pickle_dump(curr_production_model, params[\"production_model_path\"])\n",
    "    \n",
    "    # Return current chosen production model, log of production models and current training log\n",
    "    return curr_production_model, production_model_log, training_log"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 3.3.6 Save Best model and Update Training Logs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2023-01-07 22:00:43.383530 Choosing model by metrics score.\n",
      "2023-01-07 22:00:43.383530 Converting training log type of data from dict to dataframe.\n",
      "2023-01-07 22:00:43.391982 Trying to load previous production model.\n",
      "2023-01-07 22:00:43.421635 Previous production model loaded.\n",
      "2023-01-07 22:00:43.422635 Loading validation data.\n",
      "2023-01-07 22:00:43.422635 Checking compatibilty previous production model's input with current train data's features.\n",
      "2023-01-07 22:00:43.422635 Features compatible.\n",
      "2023-01-07 22:00:43.422635 Reassesing previous model performance using current validation data.\n",
      "2023-01-07 22:00:43.448835 Assessing complete.\n",
      "2023-01-07 22:00:43.449838 Storing new metrics data to previous model structure.\n",
      "2023-01-07 22:00:43.449838 Adding previous model data to current training log and list of model\n",
      "2023-01-07 22:00:43.464590 Sorting training log by f1 macro avg and training time.\n",
      "2023-01-07 22:00:43.472909 Searching model data based on sorted training log.\n",
      "2023-01-07 22:00:43.507785 Model chosen.\n"
     ]
    }
   ],
   "source": [
    "model, production_model_log, training_logs = get_production_model(list_of_trained_model, training_log, config_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 3.3.7 BASELINE MODEL"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model_name</th>\n",
       "      <th>performance</th>\n",
       "      <th>f1_score_avg</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Production-GradientBoostingClassifier</td>\n",
       "      <td>{'0': {'precision': 0.9672131147540983, 'recall': 0.7866666666666666, 'f1-score': 0.8676470588235293, 'support': 75}, '1': {'precision': 0.5897435897435898, 'recall': 0.92, 'f1-score': 0.71875, 'support': 25}, 'accuracy': 0.82, 'macro avg': {'precision': 0.7784783522488441, 'recall': 0.8533333333333333, 'f1-score': 0.7931985294117647, 'support': 100}, 'weighted avg': {'precision': 0.8728457335014713, 'recall': 0.82, 'f1-score': 0.830422794117647, 'support': 100}}</td>\n",
       "      <td>0.820000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>-oversampling-GradientBoostingClassifier</td>\n",
       "      <td>{'0': {'precision': 0.9230769230769231, 'recall': 0.8, 'f1-score': 0.8571428571428571, 'support': 75}, '1': {'precision': 0.5714285714285714, 'recall': 0.8, 'f1-score': 0.6666666666666666, 'support': 25}, 'accuracy': 0.8, 'macro avg': {'precision': 0.7472527472527473, 'recall': 0.8, 'f1-score': 0.7619047619047619, 'support': 100}, 'weighted avg': {'precision': 0.8351648351648353, 'recall': 0.8, 'f1-score': 0.8095238095238093, 'support': 100}}</td>\n",
       "      <td>0.809524</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>-nonbalance-GradientBoostingClassifier</td>\n",
       "      <td>{'0': {'precision': 0.8767123287671232, 'recall': 0.8533333333333334, 'f1-score': 0.8648648648648648, 'support': 75}, '1': {'precision': 0.5925925925925926, 'recall': 0.64, 'f1-score': 0.6153846153846153, 'support': 25}, 'accuracy': 0.8, 'macro avg': {'precision': 0.7346524606798579, 'recall': 0.7466666666666667, 'f1-score': 0.74012474012474, 'support': 100}, 'weighted avg': {'precision': 0.8056823947234905, 'recall': 0.8, 'f1-score': 0.8024948024948024, 'support': 100}}</td>\n",
       "      <td>0.802495</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>-smote-XGBClassifier</td>\n",
       "      <td>{'0': {'precision': 0.8857142857142857, 'recall': 0.8266666666666667, 'f1-score': 0.8551724137931035, 'support': 75}, '1': {'precision': 0.5666666666666667, 'recall': 0.68, 'f1-score': 0.6181818181818183, 'support': 25}, 'accuracy': 0.79, 'macro avg': {'precision': 0.7261904761904762, 'recall': 0.7533333333333334, 'f1-score': 0.7366771159874609, 'support': 100}, 'weighted avg': {'precision': 0.805952380952381, 'recall': 0.79, 'f1-score': 0.7959247648902822, 'support': 100}}</td>\n",
       "      <td>0.795925</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>-smote-DecisionTreeClassifier</td>\n",
       "      <td>{'0': {'precision': 0.8840579710144928, 'recall': 0.8133333333333334, 'f1-score': 0.8472222222222223, 'support': 75}, '1': {'precision': 0.5483870967741935, 'recall': 0.68, 'f1-score': 0.6071428571428571, 'support': 25}, 'accuracy': 0.78, 'macro avg': {'precision': 0.7162225338943431, 'recall': 0.7466666666666667, 'f1-score': 0.7271825396825398, 'support': 100}, 'weighted avg': {'precision': 0.8001402524544179, 'recall': 0.78, 'f1-score': 0.787202380952381, 'support': 100}}</td>\n",
       "      <td>0.787202</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>-smote-GradientBoostingClassifier</td>\n",
       "      <td>{'0': {'precision': 0.8840579710144928, 'recall': 0.8133333333333334, 'f1-score': 0.8472222222222223, 'support': 75}, '1': {'precision': 0.5483870967741935, 'recall': 0.68, 'f1-score': 0.6071428571428571, 'support': 25}, 'accuracy': 0.78, 'macro avg': {'precision': 0.7162225338943431, 'recall': 0.7466666666666667, 'f1-score': 0.7271825396825398, 'support': 100}, 'weighted avg': {'precision': 0.8001402524544179, 'recall': 0.78, 'f1-score': 0.787202380952381, 'support': 100}}</td>\n",
       "      <td>0.787202</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>-oversampling-XGBClassifier</td>\n",
       "      <td>{'0': {'precision': 0.8714285714285714, 'recall': 0.8133333333333334, 'f1-score': 0.8413793103448276, 'support': 75}, '1': {'precision': 0.5333333333333333, 'recall': 0.64, 'f1-score': 0.5818181818181818, 'support': 25}, 'accuracy': 0.77, 'macro avg': {'precision': 0.7023809523809523, 'recall': 0.7266666666666667, 'f1-score': 0.7115987460815048, 'support': 100}, 'weighted avg': {'precision': 0.7869047619047619, 'recall': 0.77, 'f1-score': 0.7764890282131661, 'support': 100}}</td>\n",
       "      <td>0.776489</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>-nonbalance-DecisionTreeClassifier</td>\n",
       "      <td>{'0': {'precision': 0.8513513513513513, 'recall': 0.84, 'f1-score': 0.8456375838926175, 'support': 75}, '1': {'precision': 0.5384615384615384, 'recall': 0.56, 'f1-score': 0.5490196078431373, 'support': 25}, 'accuracy': 0.77, 'macro avg': {'precision': 0.6949064449064448, 'recall': 0.7, 'f1-score': 0.6973285958678774, 'support': 100}, 'weighted avg': {'precision': 0.7731288981288981, 'recall': 0.77, 'f1-score': 0.7714830898802474, 'support': 100}}</td>\n",
       "      <td>0.771483</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>-oversampling-DecisionTreeClassifier</td>\n",
       "      <td>{'0': {'precision': 0.8591549295774648, 'recall': 0.8133333333333334, 'f1-score': 0.8356164383561644, 'support': 75}, '1': {'precision': 0.5172413793103449, 'recall': 0.6, 'f1-score': 0.5555555555555556, 'support': 25}, 'accuracy': 0.76, 'macro avg': {'precision': 0.6881981544439049, 'recall': 0.7066666666666667, 'f1-score': 0.69558599695586, 'support': 100}, 'weighted avg': {'precision': 0.7736765420106848, 'recall': 0.76, 'f1-score': 0.7656012176560122, 'support': 100}}</td>\n",
       "      <td>0.765601</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>-nonbalance-LogisticRegression</td>\n",
       "      <td>{'0': {'precision': 0.8192771084337349, 'recall': 0.9066666666666666, 'f1-score': 0.8607594936708859, 'support': 75}, '1': {'precision': 0.5882352941176471, 'recall': 0.4, 'f1-score': 0.4761904761904762, 'support': 25}, 'accuracy': 0.78, 'macro avg': {'precision': 0.703756201275691, 'recall': 0.6533333333333333, 'f1-score': 0.668474984930681, 'support': 100}, 'weighted avg': {'precision': 0.761516654854713, 'recall': 0.78, 'f1-score': 0.7646172393007834, 'support': 100}}</td>\n",
       "      <td>0.764617</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>-oversampling-AdaBoostClassifier</td>\n",
       "      <td>{'0': {'precision': 0.8493150684931506, 'recall': 0.8266666666666667, 'f1-score': 0.8378378378378378, 'support': 75}, '1': {'precision': 0.5185185185185185, 'recall': 0.56, 'f1-score': 0.5384615384615384, 'support': 25}, 'accuracy': 0.76, 'macro avg': {'precision': 0.6839167935058346, 'recall': 0.6933333333333334, 'f1-score': 0.6881496881496881, 'support': 100}, 'weighted avg': {'precision': 0.7666159309994925, 'recall': 0.76, 'f1-score': 0.7629937629937631, 'support': 100}}</td>\n",
       "      <td>0.762994</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>-nonbalance-XGBClassifier</td>\n",
       "      <td>{'0': {'precision': 0.84, 'recall': 0.84, 'f1-score': 0.8399999999999999, 'support': 75}, '1': {'precision': 0.52, 'recall': 0.52, 'f1-score': 0.52, 'support': 25}, 'accuracy': 0.76, 'macro avg': {'precision': 0.6799999999999999, 'recall': 0.6799999999999999, 'f1-score': 0.6799999999999999, 'support': 100}, 'weighted avg': {'precision': 0.76, 'recall': 0.76, 'f1-score': 0.7599999999999999, 'support': 100}}</td>\n",
       "      <td>0.760000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>-nonbalance-AdaBoostClassifier</td>\n",
       "      <td>{'0': {'precision': 0.8205128205128205, 'recall': 0.8533333333333334, 'f1-score': 0.8366013071895425, 'support': 75}, '1': {'precision': 0.5, 'recall': 0.44, 'f1-score': 0.46808510638297873, 'support': 25}, 'accuracy': 0.75, 'macro avg': {'precision': 0.6602564102564102, 'recall': 0.6466666666666667, 'f1-score': 0.6523432067862607, 'support': 100}, 'weighted avg': {'precision': 0.7403846153846153, 'recall': 0.75, 'f1-score': 0.7444722569879016, 'support': 100}}</td>\n",
       "      <td>0.744472</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>-smote-LogisticRegression</td>\n",
       "      <td>{'0': {'precision': 0.8529411764705882, 'recall': 0.7733333333333333, 'f1-score': 0.8111888111888111, 'support': 75}, '1': {'precision': 0.46875, 'recall': 0.6, 'f1-score': 0.5263157894736842, 'support': 25}, 'accuracy': 0.73, 'macro avg': {'precision': 0.6608455882352942, 'recall': 0.6866666666666666, 'f1-score': 0.6687523003312477, 'support': 100}, 'weighted avg': {'precision': 0.7568933823529411, 'recall': 0.73, 'f1-score': 0.7399705557600293, 'support': 100}}</td>\n",
       "      <td>0.739971</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>-oversampling-LogisticRegression</td>\n",
       "      <td>{'0': {'precision': 0.8615384615384616, 'recall': 0.7466666666666667, 'f1-score': 0.8, 'support': 75}, '1': {'precision': 0.45714285714285713, 'recall': 0.64, 'f1-score': 0.5333333333333333, 'support': 25}, 'accuracy': 0.72, 'macro avg': {'precision': 0.6593406593406593, 'recall': 0.6933333333333334, 'f1-score': 0.6666666666666667, 'support': 100}, 'weighted avg': {'precision': 0.7604395604395604, 'recall': 0.72, 'f1-score': 0.7333333333333333, 'support': 100}}</td>\n",
       "      <td>0.733333</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>-oversampling-RandomForestClassifier</td>\n",
       "      <td>{'0': {'precision': 0.7976190476190477, 'recall': 0.8933333333333333, 'f1-score': 0.8427672955974843, 'support': 75}, '1': {'precision': 0.5, 'recall': 0.32, 'f1-score': 0.39024390243902435, 'support': 25}, 'accuracy': 0.75, 'macro avg': {'precision': 0.6488095238095238, 'recall': 0.6066666666666667, 'f1-score': 0.6165055990182543, 'support': 100}, 'weighted avg': {'precision': 0.7232142857142858, 'recall': 0.75, 'f1-score': 0.7296364473078693, 'support': 100}}</td>\n",
       "      <td>0.729636</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>-nonbalance-SVC</td>\n",
       "      <td>{'0': {'precision': 0.7802197802197802, 'recall': 0.9466666666666667, 'f1-score': 0.8554216867469879, 'support': 75}, '1': {'precision': 0.5555555555555556, 'recall': 0.2, 'f1-score': 0.29411764705882354, 'support': 25}, 'accuracy': 0.76, 'macro avg': {'precision': 0.667887667887668, 'recall': 0.5733333333333334, 'f1-score': 0.5747696669029058, 'support': 100}, 'weighted avg': {'precision': 0.724053724053724, 'recall': 0.76, 'f1-score': 0.7150956768249469, 'support': 100}}</td>\n",
       "      <td>0.715096</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>-smote-AdaBoostClassifier</td>\n",
       "      <td>{'0': {'precision': 0.7974683544303798, 'recall': 0.84, 'f1-score': 0.8181818181818181, 'support': 75}, '1': {'precision': 0.42857142857142855, 'recall': 0.36, 'f1-score': 0.391304347826087, 'support': 25}, 'accuracy': 0.72, 'macro avg': {'precision': 0.6130198915009042, 'recall': 0.6, 'f1-score': 0.6047430830039525, 'support': 100}, 'weighted avg': {'precision': 0.705244122965642, 'recall': 0.72, 'f1-score': 0.7114624505928853, 'support': 100}}</td>\n",
       "      <td>0.711462</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>-oversampling-SVC</td>\n",
       "      <td>{'0': {'precision': 0.7901234567901234, 'recall': 0.8533333333333334, 'f1-score': 0.8205128205128205, 'support': 75}, '1': {'precision': 0.42105263157894735, 'recall': 0.32, 'f1-score': 0.3636363636363636, 'support': 25}, 'accuracy': 0.72, 'macro avg': {'precision': 0.6055880441845354, 'recall': 0.5866666666666667, 'f1-score': 0.592074592074592, 'support': 100}, 'weighted avg': {'precision': 0.6978557504873294, 'recall': 0.72, 'f1-score': 0.7062937062937062, 'support': 100}}</td>\n",
       "      <td>0.706294</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>-smote-RandomForestClassifier</td>\n",
       "      <td>{'0': {'precision': 0.7752808988764045, 'recall': 0.92, 'f1-score': 0.8414634146341463, 'support': 75}, '1': {'precision': 0.45454545454545453, 'recall': 0.2, 'f1-score': 0.2777777777777778, 'support': 25}, 'accuracy': 0.74, 'macro avg': {'precision': 0.6149131767109295, 'recall': 0.56, 'f1-score': 0.559620596205962, 'support': 100}, 'weighted avg': {'precision': 0.695097037793667, 'recall': 0.74, 'f1-score': 0.7005420054200542, 'support': 100}}</td>\n",
       "      <td>0.700542</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>-smote-GaussianNB</td>\n",
       "      <td>{'0': {'precision': 0.85, 'recall': 0.68, 'f1-score': 0.7555555555555556, 'support': 75}, '1': {'precision': 0.4, 'recall': 0.64, 'f1-score': 0.4923076923076923, 'support': 25}, 'accuracy': 0.67, 'macro avg': {'precision': 0.625, 'recall': 0.66, 'f1-score': 0.6239316239316239, 'support': 100}, 'weighted avg': {'precision': 0.7375, 'recall': 0.67, 'f1-score': 0.6897435897435897, 'support': 100}}</td>\n",
       "      <td>0.689744</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>-smote-SVC</td>\n",
       "      <td>{'0': {'precision': 0.7674418604651163, 'recall': 0.88, 'f1-score': 0.8198757763975156, 'support': 75}, '1': {'precision': 0.35714285714285715, 'recall': 0.2, 'f1-score': 0.25641025641025644, 'support': 25}, 'accuracy': 0.71, 'macro avg': {'precision': 0.5622923588039868, 'recall': 0.54, 'f1-score': 0.5381430164038861, 'support': 100}, 'weighted avg': {'precision': 0.6648671096345515, 'recall': 0.71, 'f1-score': 0.6790093964007008, 'support': 100}}</td>\n",
       "      <td>0.679009</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>-nonbalance-GaussianNB</td>\n",
       "      <td>{'0': {'precision': 0.8448275862068966, 'recall': 0.6533333333333333, 'f1-score': 0.7368421052631579, 'support': 75}, '1': {'precision': 0.38095238095238093, 'recall': 0.64, 'f1-score': 0.4776119402985074, 'support': 25}, 'accuracy': 0.65, 'macro avg': {'precision': 0.6128899835796388, 'recall': 0.6466666666666667, 'f1-score': 0.6072270227808326, 'support': 100}, 'weighted avg': {'precision': 0.7288587848932677, 'recall': 0.65, 'f1-score': 0.6720345640219952, 'support': 100}}</td>\n",
       "      <td>0.672035</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>-nonbalance-RandomForestClassifier</td>\n",
       "      <td>{'0': {'precision': 0.7578947368421053, 'recall': 0.96, 'f1-score': 0.8470588235294118, 'support': 75}, '1': {'precision': 0.4, 'recall': 0.08, 'f1-score': 0.13333333333333333, 'support': 25}, 'accuracy': 0.74, 'macro avg': {'precision': 0.5789473684210527, 'recall': 0.52, 'f1-score': 0.49019607843137253, 'support': 100}, 'weighted avg': {'precision': 0.6684210526315789, 'recall': 0.74, 'f1-score': 0.6686274509803921, 'support': 100}}</td>\n",
       "      <td>0.668627</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>-oversampling-QuadraticDiscriminantAnalysis</td>\n",
       "      <td>{'0': {'precision': 0.7578947368421053, 'recall': 0.96, 'f1-score': 0.8470588235294118, 'support': 75}, '1': {'precision': 0.4, 'recall': 0.08, 'f1-score': 0.13333333333333333, 'support': 25}, 'accuracy': 0.74, 'macro avg': {'precision': 0.5789473684210527, 'recall': 0.52, 'f1-score': 0.49019607843137253, 'support': 100}, 'weighted avg': {'precision': 0.6684210526315789, 'recall': 0.74, 'f1-score': 0.6686274509803921, 'support': 100}}</td>\n",
       "      <td>0.668627</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>-oversampling-GaussianNB</td>\n",
       "      <td>{'0': {'precision': 0.819672131147541, 'recall': 0.6666666666666666, 'f1-score': 0.7352941176470588, 'support': 75}, '1': {'precision': 0.358974358974359, 'recall': 0.56, 'f1-score': 0.4375, 'support': 25}, 'accuracy': 0.64, 'macro avg': {'precision': 0.58932324506095, 'recall': 0.6133333333333333, 'f1-score': 0.5863970588235294, 'support': 100}, 'weighted avg': {'precision': 0.7044976881042454, 'recall': 0.64, 'f1-score': 0.660845588235294, 'support': 100}}</td>\n",
       "      <td>0.660846</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>-smote-QuadraticDiscriminantAnalysis</td>\n",
       "      <td>{'0': {'precision': 0.7473684210526316, 'recall': 0.9466666666666667, 'f1-score': 0.8352941176470587, 'support': 75}, '1': {'precision': 0.2, 'recall': 0.04, 'f1-score': 0.06666666666666667, 'support': 25}, 'accuracy': 0.72, 'macro avg': {'precision': 0.4736842105263158, 'recall': 0.49333333333333335, 'f1-score': 0.4509803921568627, 'support': 100}, 'weighted avg': {'precision': 0.6105263157894737, 'recall': 0.72, 'f1-score': 0.6431372549019607, 'support': 100}}</td>\n",
       "      <td>0.643137</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>-nonbalance-KNeighborsClassifier</td>\n",
       "      <td>{'0': {'precision': 0.7444444444444445, 'recall': 0.8933333333333333, 'f1-score': 0.8121212121212122, 'support': 75}, '1': {'precision': 0.2, 'recall': 0.08, 'f1-score': 0.11428571428571428, 'support': 25}, 'accuracy': 0.69, 'macro avg': {'precision': 0.4722222222222222, 'recall': 0.48666666666666664, 'f1-score': 0.4632034632034633, 'support': 100}, 'weighted avg': {'precision': 0.6083333333333334, 'recall': 0.69, 'f1-score': 0.6376623376623377, 'support': 100}}</td>\n",
       "      <td>0.637662</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>-nonbalance-QuadraticDiscriminantAnalysis</td>\n",
       "      <td>{'0': {'precision': 0.7160493827160493, 'recall': 0.7733333333333333, 'f1-score': 0.7435897435897435, 'support': 75}, '1': {'precision': 0.10526315789473684, 'recall': 0.08, 'f1-score': 0.0909090909090909, 'support': 25}, 'accuracy': 0.6, 'macro avg': {'precision': 0.4106562703053931, 'recall': 0.42666666666666664, 'f1-score': 0.4172494172494172, 'support': 100}, 'weighted avg': {'precision': 0.5633528265107212, 'recall': 0.6, 'f1-score': 0.5804195804195803, 'support': 100}}</td>\n",
       "      <td>0.580420</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>-oversampling-KNeighborsClassifier</td>\n",
       "      <td>{'0': {'precision': 0.7457627118644068, 'recall': 0.5866666666666667, 'f1-score': 0.6567164179104478, 'support': 75}, '1': {'precision': 0.24390243902439024, 'recall': 0.4, 'f1-score': 0.30303030303030304, 'support': 25}, 'accuracy': 0.54, 'macro avg': {'precision': 0.4948325754443985, 'recall': 0.49333333333333335, 'f1-score': 0.47987336047037543, 'support': 100}, 'weighted avg': {'precision': 0.6202976436544028, 'recall': 0.54, 'f1-score': 0.5682948891904116, 'support': 100}}</td>\n",
       "      <td>0.568295</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>-smote-KNeighborsClassifier</td>\n",
       "      <td>{'0': {'precision': 0.9230769230769231, 'recall': 0.32, 'f1-score': 0.4752475247524752, 'support': 75}, '1': {'precision': 0.3108108108108108, 'recall': 0.92, 'f1-score': 0.46464646464646464, 'support': 25}, 'accuracy': 0.47, 'macro avg': {'precision': 0.6169438669438669, 'recall': 0.62, 'f1-score': 0.46994699469946993, 'support': 100}, 'weighted avg': {'precision': 0.7700103950103951, 'recall': 0.47, 'f1-score': 0.4725972597259725, 'support': 100}}</td>\n",
       "      <td>0.472597</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                     model_name  \\\n",
       "0         Production-GradientBoostingClassifier   \n",
       "26     -oversampling-GradientBoostingClassifier   \n",
       "6        -nonbalance-GradientBoostingClassifier   \n",
       "17                         -smote-XGBClassifier   \n",
       "11                -smote-DecisionTreeClassifier   \n",
       "16            -smote-GradientBoostingClassifier   \n",
       "27                  -oversampling-XGBClassifier   \n",
       "1            -nonbalance-DecisionTreeClassifier   \n",
       "21         -oversampling-DecisionTreeClassifier   \n",
       "2                -nonbalance-LogisticRegression   \n",
       "25             -oversampling-AdaBoostClassifier   \n",
       "7                     -nonbalance-XGBClassifier   \n",
       "5                -nonbalance-AdaBoostClassifier   \n",
       "12                    -smote-LogisticRegression   \n",
       "22             -oversampling-LogisticRegression   \n",
       "24         -oversampling-RandomForestClassifier   \n",
       "3                               -nonbalance-SVC   \n",
       "15                    -smote-AdaBoostClassifier   \n",
       "23                            -oversampling-SVC   \n",
       "14                -smote-RandomForestClassifier   \n",
       "19                            -smote-GaussianNB   \n",
       "13                                   -smote-SVC   \n",
       "9                        -nonbalance-GaussianNB   \n",
       "4            -nonbalance-RandomForestClassifier   \n",
       "28  -oversampling-QuadraticDiscriminantAnalysis   \n",
       "29                     -oversampling-GaussianNB   \n",
       "18         -smote-QuadraticDiscriminantAnalysis   \n",
       "0              -nonbalance-KNeighborsClassifier   \n",
       "8     -nonbalance-QuadraticDiscriminantAnalysis   \n",
       "20           -oversampling-KNeighborsClassifier   \n",
       "10                  -smote-KNeighborsClassifier   \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                                                                                                                                                                                                           performance  \\\n",
       "0                  {'0': {'precision': 0.9672131147540983, 'recall': 0.7866666666666666, 'f1-score': 0.8676470588235293, 'support': 75}, '1': {'precision': 0.5897435897435898, 'recall': 0.92, 'f1-score': 0.71875, 'support': 25}, 'accuracy': 0.82, 'macro avg': {'precision': 0.7784783522488441, 'recall': 0.8533333333333333, 'f1-score': 0.7931985294117647, 'support': 100}, 'weighted avg': {'precision': 0.8728457335014713, 'recall': 0.82, 'f1-score': 0.830422794117647, 'support': 100}}   \n",
       "26                                      {'0': {'precision': 0.9230769230769231, 'recall': 0.8, 'f1-score': 0.8571428571428571, 'support': 75}, '1': {'precision': 0.5714285714285714, 'recall': 0.8, 'f1-score': 0.6666666666666666, 'support': 25}, 'accuracy': 0.8, 'macro avg': {'precision': 0.7472527472527473, 'recall': 0.8, 'f1-score': 0.7619047619047619, 'support': 100}, 'weighted avg': {'precision': 0.8351648351648353, 'recall': 0.8, 'f1-score': 0.8095238095238093, 'support': 100}}   \n",
       "6          {'0': {'precision': 0.8767123287671232, 'recall': 0.8533333333333334, 'f1-score': 0.8648648648648648, 'support': 75}, '1': {'precision': 0.5925925925925926, 'recall': 0.64, 'f1-score': 0.6153846153846153, 'support': 25}, 'accuracy': 0.8, 'macro avg': {'precision': 0.7346524606798579, 'recall': 0.7466666666666667, 'f1-score': 0.74012474012474, 'support': 100}, 'weighted avg': {'precision': 0.8056823947234905, 'recall': 0.8, 'f1-score': 0.8024948024948024, 'support': 100}}   \n",
       "17      {'0': {'precision': 0.8857142857142857, 'recall': 0.8266666666666667, 'f1-score': 0.8551724137931035, 'support': 75}, '1': {'precision': 0.5666666666666667, 'recall': 0.68, 'f1-score': 0.6181818181818183, 'support': 25}, 'accuracy': 0.79, 'macro avg': {'precision': 0.7261904761904762, 'recall': 0.7533333333333334, 'f1-score': 0.7366771159874609, 'support': 100}, 'weighted avg': {'precision': 0.805952380952381, 'recall': 0.79, 'f1-score': 0.7959247648902822, 'support': 100}}   \n",
       "11      {'0': {'precision': 0.8840579710144928, 'recall': 0.8133333333333334, 'f1-score': 0.8472222222222223, 'support': 75}, '1': {'precision': 0.5483870967741935, 'recall': 0.68, 'f1-score': 0.6071428571428571, 'support': 25}, 'accuracy': 0.78, 'macro avg': {'precision': 0.7162225338943431, 'recall': 0.7466666666666667, 'f1-score': 0.7271825396825398, 'support': 100}, 'weighted avg': {'precision': 0.8001402524544179, 'recall': 0.78, 'f1-score': 0.787202380952381, 'support': 100}}   \n",
       "16      {'0': {'precision': 0.8840579710144928, 'recall': 0.8133333333333334, 'f1-score': 0.8472222222222223, 'support': 75}, '1': {'precision': 0.5483870967741935, 'recall': 0.68, 'f1-score': 0.6071428571428571, 'support': 25}, 'accuracy': 0.78, 'macro avg': {'precision': 0.7162225338943431, 'recall': 0.7466666666666667, 'f1-score': 0.7271825396825398, 'support': 100}, 'weighted avg': {'precision': 0.8001402524544179, 'recall': 0.78, 'f1-score': 0.787202380952381, 'support': 100}}   \n",
       "27     {'0': {'precision': 0.8714285714285714, 'recall': 0.8133333333333334, 'f1-score': 0.8413793103448276, 'support': 75}, '1': {'precision': 0.5333333333333333, 'recall': 0.64, 'f1-score': 0.5818181818181818, 'support': 25}, 'accuracy': 0.77, 'macro avg': {'precision': 0.7023809523809523, 'recall': 0.7266666666666667, 'f1-score': 0.7115987460815048, 'support': 100}, 'weighted avg': {'precision': 0.7869047619047619, 'recall': 0.77, 'f1-score': 0.7764890282131661, 'support': 100}}   \n",
       "1                                   {'0': {'precision': 0.8513513513513513, 'recall': 0.84, 'f1-score': 0.8456375838926175, 'support': 75}, '1': {'precision': 0.5384615384615384, 'recall': 0.56, 'f1-score': 0.5490196078431373, 'support': 25}, 'accuracy': 0.77, 'macro avg': {'precision': 0.6949064449064448, 'recall': 0.7, 'f1-score': 0.6973285958678774, 'support': 100}, 'weighted avg': {'precision': 0.7731288981288981, 'recall': 0.77, 'f1-score': 0.7714830898802474, 'support': 100}}   \n",
       "21        {'0': {'precision': 0.8591549295774648, 'recall': 0.8133333333333334, 'f1-score': 0.8356164383561644, 'support': 75}, '1': {'precision': 0.5172413793103449, 'recall': 0.6, 'f1-score': 0.5555555555555556, 'support': 25}, 'accuracy': 0.76, 'macro avg': {'precision': 0.6881981544439049, 'recall': 0.7066666666666667, 'f1-score': 0.69558599695586, 'support': 100}, 'weighted avg': {'precision': 0.7736765420106848, 'recall': 0.76, 'f1-score': 0.7656012176560122, 'support': 100}}   \n",
       "2          {'0': {'precision': 0.8192771084337349, 'recall': 0.9066666666666666, 'f1-score': 0.8607594936708859, 'support': 75}, '1': {'precision': 0.5882352941176471, 'recall': 0.4, 'f1-score': 0.4761904761904762, 'support': 25}, 'accuracy': 0.78, 'macro avg': {'precision': 0.703756201275691, 'recall': 0.6533333333333333, 'f1-score': 0.668474984930681, 'support': 100}, 'weighted avg': {'precision': 0.761516654854713, 'recall': 0.78, 'f1-score': 0.7646172393007834, 'support': 100}}   \n",
       "25     {'0': {'precision': 0.8493150684931506, 'recall': 0.8266666666666667, 'f1-score': 0.8378378378378378, 'support': 75}, '1': {'precision': 0.5185185185185185, 'recall': 0.56, 'f1-score': 0.5384615384615384, 'support': 25}, 'accuracy': 0.76, 'macro avg': {'precision': 0.6839167935058346, 'recall': 0.6933333333333334, 'f1-score': 0.6881496881496881, 'support': 100}, 'weighted avg': {'precision': 0.7666159309994925, 'recall': 0.76, 'f1-score': 0.7629937629937631, 'support': 100}}   \n",
       "7                                                                            {'0': {'precision': 0.84, 'recall': 0.84, 'f1-score': 0.8399999999999999, 'support': 75}, '1': {'precision': 0.52, 'recall': 0.52, 'f1-score': 0.52, 'support': 25}, 'accuracy': 0.76, 'macro avg': {'precision': 0.6799999999999999, 'recall': 0.6799999999999999, 'f1-score': 0.6799999999999999, 'support': 100}, 'weighted avg': {'precision': 0.76, 'recall': 0.76, 'f1-score': 0.7599999999999999, 'support': 100}}   \n",
       "5                    {'0': {'precision': 0.8205128205128205, 'recall': 0.8533333333333334, 'f1-score': 0.8366013071895425, 'support': 75}, '1': {'precision': 0.5, 'recall': 0.44, 'f1-score': 0.46808510638297873, 'support': 25}, 'accuracy': 0.75, 'macro avg': {'precision': 0.6602564102564102, 'recall': 0.6466666666666667, 'f1-score': 0.6523432067862607, 'support': 100}, 'weighted avg': {'precision': 0.7403846153846153, 'recall': 0.75, 'f1-score': 0.7444722569879016, 'support': 100}}   \n",
       "12                 {'0': {'precision': 0.8529411764705882, 'recall': 0.7733333333333333, 'f1-score': 0.8111888111888111, 'support': 75}, '1': {'precision': 0.46875, 'recall': 0.6, 'f1-score': 0.5263157894736842, 'support': 25}, 'accuracy': 0.73, 'macro avg': {'precision': 0.6608455882352942, 'recall': 0.6866666666666666, 'f1-score': 0.6687523003312477, 'support': 100}, 'weighted avg': {'precision': 0.7568933823529411, 'recall': 0.73, 'f1-score': 0.7399705557600293, 'support': 100}}   \n",
       "22                   {'0': {'precision': 0.8615384615384616, 'recall': 0.7466666666666667, 'f1-score': 0.8, 'support': 75}, '1': {'precision': 0.45714285714285713, 'recall': 0.64, 'f1-score': 0.5333333333333333, 'support': 25}, 'accuracy': 0.72, 'macro avg': {'precision': 0.6593406593406593, 'recall': 0.6933333333333334, 'f1-score': 0.6666666666666667, 'support': 100}, 'weighted avg': {'precision': 0.7604395604395604, 'recall': 0.72, 'f1-score': 0.7333333333333333, 'support': 100}}   \n",
       "24                   {'0': {'precision': 0.7976190476190477, 'recall': 0.8933333333333333, 'f1-score': 0.8427672955974843, 'support': 75}, '1': {'precision': 0.5, 'recall': 0.32, 'f1-score': 0.39024390243902435, 'support': 25}, 'accuracy': 0.75, 'macro avg': {'precision': 0.6488095238095238, 'recall': 0.6066666666666667, 'f1-score': 0.6165055990182543, 'support': 100}, 'weighted avg': {'precision': 0.7232142857142858, 'recall': 0.75, 'f1-score': 0.7296364473078693, 'support': 100}}   \n",
       "3        {'0': {'precision': 0.7802197802197802, 'recall': 0.9466666666666667, 'f1-score': 0.8554216867469879, 'support': 75}, '1': {'precision': 0.5555555555555556, 'recall': 0.2, 'f1-score': 0.29411764705882354, 'support': 25}, 'accuracy': 0.76, 'macro avg': {'precision': 0.667887667887668, 'recall': 0.5733333333333334, 'f1-score': 0.5747696669029058, 'support': 100}, 'weighted avg': {'precision': 0.724053724053724, 'recall': 0.76, 'f1-score': 0.7150956768249469, 'support': 100}}   \n",
       "15                                   {'0': {'precision': 0.7974683544303798, 'recall': 0.84, 'f1-score': 0.8181818181818181, 'support': 75}, '1': {'precision': 0.42857142857142855, 'recall': 0.36, 'f1-score': 0.391304347826087, 'support': 25}, 'accuracy': 0.72, 'macro avg': {'precision': 0.6130198915009042, 'recall': 0.6, 'f1-score': 0.6047430830039525, 'support': 100}, 'weighted avg': {'precision': 0.705244122965642, 'recall': 0.72, 'f1-score': 0.7114624505928853, 'support': 100}}   \n",
       "23     {'0': {'precision': 0.7901234567901234, 'recall': 0.8533333333333334, 'f1-score': 0.8205128205128205, 'support': 75}, '1': {'precision': 0.42105263157894735, 'recall': 0.32, 'f1-score': 0.3636363636363636, 'support': 25}, 'accuracy': 0.72, 'macro avg': {'precision': 0.6055880441845354, 'recall': 0.5866666666666667, 'f1-score': 0.592074592074592, 'support': 100}, 'weighted avg': {'precision': 0.6978557504873294, 'recall': 0.72, 'f1-score': 0.7062937062937062, 'support': 100}}   \n",
       "14                                   {'0': {'precision': 0.7752808988764045, 'recall': 0.92, 'f1-score': 0.8414634146341463, 'support': 75}, '1': {'precision': 0.45454545454545453, 'recall': 0.2, 'f1-score': 0.2777777777777778, 'support': 25}, 'accuracy': 0.74, 'macro avg': {'precision': 0.6149131767109295, 'recall': 0.56, 'f1-score': 0.559620596205962, 'support': 100}, 'weighted avg': {'precision': 0.695097037793667, 'recall': 0.74, 'f1-score': 0.7005420054200542, 'support': 100}}   \n",
       "19                                                                                       {'0': {'precision': 0.85, 'recall': 0.68, 'f1-score': 0.7555555555555556, 'support': 75}, '1': {'precision': 0.4, 'recall': 0.64, 'f1-score': 0.4923076923076923, 'support': 25}, 'accuracy': 0.67, 'macro avg': {'precision': 0.625, 'recall': 0.66, 'f1-score': 0.6239316239316239, 'support': 100}, 'weighted avg': {'precision': 0.7375, 'recall': 0.67, 'f1-score': 0.6897435897435897, 'support': 100}}   \n",
       "13                                {'0': {'precision': 0.7674418604651163, 'recall': 0.88, 'f1-score': 0.8198757763975156, 'support': 75}, '1': {'precision': 0.35714285714285715, 'recall': 0.2, 'f1-score': 0.25641025641025644, 'support': 25}, 'accuracy': 0.71, 'macro avg': {'precision': 0.5622923588039868, 'recall': 0.54, 'f1-score': 0.5381430164038861, 'support': 100}, 'weighted avg': {'precision': 0.6648671096345515, 'recall': 0.71, 'f1-score': 0.6790093964007008, 'support': 100}}   \n",
       "9     {'0': {'precision': 0.8448275862068966, 'recall': 0.6533333333333333, 'f1-score': 0.7368421052631579, 'support': 75}, '1': {'precision': 0.38095238095238093, 'recall': 0.64, 'f1-score': 0.4776119402985074, 'support': 25}, 'accuracy': 0.65, 'macro avg': {'precision': 0.6128899835796388, 'recall': 0.6466666666666667, 'f1-score': 0.6072270227808326, 'support': 100}, 'weighted avg': {'precision': 0.7288587848932677, 'recall': 0.65, 'f1-score': 0.6720345640219952, 'support': 100}}   \n",
       "4                                               {'0': {'precision': 0.7578947368421053, 'recall': 0.96, 'f1-score': 0.8470588235294118, 'support': 75}, '1': {'precision': 0.4, 'recall': 0.08, 'f1-score': 0.13333333333333333, 'support': 25}, 'accuracy': 0.74, 'macro avg': {'precision': 0.5789473684210527, 'recall': 0.52, 'f1-score': 0.49019607843137253, 'support': 100}, 'weighted avg': {'precision': 0.6684210526315789, 'recall': 0.74, 'f1-score': 0.6686274509803921, 'support': 100}}   \n",
       "28                                              {'0': {'precision': 0.7578947368421053, 'recall': 0.96, 'f1-score': 0.8470588235294118, 'support': 75}, '1': {'precision': 0.4, 'recall': 0.08, 'f1-score': 0.13333333333333333, 'support': 25}, 'accuracy': 0.74, 'macro avg': {'precision': 0.5789473684210527, 'recall': 0.52, 'f1-score': 0.49019607843137253, 'support': 100}, 'weighted avg': {'precision': 0.6684210526315789, 'recall': 0.74, 'f1-score': 0.6686274509803921, 'support': 100}}   \n",
       "29                      {'0': {'precision': 0.819672131147541, 'recall': 0.6666666666666666, 'f1-score': 0.7352941176470588, 'support': 75}, '1': {'precision': 0.358974358974359, 'recall': 0.56, 'f1-score': 0.4375, 'support': 25}, 'accuracy': 0.64, 'macro avg': {'precision': 0.58932324506095, 'recall': 0.6133333333333333, 'f1-score': 0.5863970588235294, 'support': 100}, 'weighted avg': {'precision': 0.7044976881042454, 'recall': 0.64, 'f1-score': 0.660845588235294, 'support': 100}}   \n",
       "18                  {'0': {'precision': 0.7473684210526316, 'recall': 0.9466666666666667, 'f1-score': 0.8352941176470587, 'support': 75}, '1': {'precision': 0.2, 'recall': 0.04, 'f1-score': 0.06666666666666667, 'support': 25}, 'accuracy': 0.72, 'macro avg': {'precision': 0.4736842105263158, 'recall': 0.49333333333333335, 'f1-score': 0.4509803921568627, 'support': 100}, 'weighted avg': {'precision': 0.6105263157894737, 'recall': 0.72, 'f1-score': 0.6431372549019607, 'support': 100}}   \n",
       "0                   {'0': {'precision': 0.7444444444444445, 'recall': 0.8933333333333333, 'f1-score': 0.8121212121212122, 'support': 75}, '1': {'precision': 0.2, 'recall': 0.08, 'f1-score': 0.11428571428571428, 'support': 25}, 'accuracy': 0.69, 'macro avg': {'precision': 0.4722222222222222, 'recall': 0.48666666666666664, 'f1-score': 0.4632034632034633, 'support': 100}, 'weighted avg': {'precision': 0.6083333333333334, 'recall': 0.69, 'f1-score': 0.6376623376623377, 'support': 100}}   \n",
       "8      {'0': {'precision': 0.7160493827160493, 'recall': 0.7733333333333333, 'f1-score': 0.7435897435897435, 'support': 75}, '1': {'precision': 0.10526315789473684, 'recall': 0.08, 'f1-score': 0.0909090909090909, 'support': 25}, 'accuracy': 0.6, 'macro avg': {'precision': 0.4106562703053931, 'recall': 0.42666666666666664, 'f1-score': 0.4172494172494172, 'support': 100}, 'weighted avg': {'precision': 0.5633528265107212, 'recall': 0.6, 'f1-score': 0.5804195804195803, 'support': 100}}   \n",
       "20  {'0': {'precision': 0.7457627118644068, 'recall': 0.5866666666666667, 'f1-score': 0.6567164179104478, 'support': 75}, '1': {'precision': 0.24390243902439024, 'recall': 0.4, 'f1-score': 0.30303030303030304, 'support': 25}, 'accuracy': 0.54, 'macro avg': {'precision': 0.4948325754443985, 'recall': 0.49333333333333335, 'f1-score': 0.47987336047037543, 'support': 100}, 'weighted avg': {'precision': 0.6202976436544028, 'recall': 0.54, 'f1-score': 0.5682948891904116, 'support': 100}}   \n",
       "10                               {'0': {'precision': 0.9230769230769231, 'recall': 0.32, 'f1-score': 0.4752475247524752, 'support': 75}, '1': {'precision': 0.3108108108108108, 'recall': 0.92, 'f1-score': 0.46464646464646464, 'support': 25}, 'accuracy': 0.47, 'macro avg': {'precision': 0.6169438669438669, 'recall': 0.62, 'f1-score': 0.46994699469946993, 'support': 100}, 'weighted avg': {'precision': 0.7700103950103951, 'recall': 0.47, 'f1-score': 0.4725972597259725, 'support': 100}}   \n",
       "\n",
       "    f1_score_avg  \n",
       "0       0.820000  \n",
       "26      0.809524  \n",
       "6       0.802495  \n",
       "17      0.795925  \n",
       "11      0.787202  \n",
       "16      0.787202  \n",
       "27      0.776489  \n",
       "1       0.771483  \n",
       "21      0.765601  \n",
       "2       0.764617  \n",
       "25      0.762994  \n",
       "7       0.760000  \n",
       "5       0.744472  \n",
       "12      0.739971  \n",
       "22      0.733333  \n",
       "24      0.729636  \n",
       "3       0.715096  \n",
       "15      0.711462  \n",
       "23      0.706294  \n",
       "14      0.700542  \n",
       "19      0.689744  \n",
       "13      0.679009  \n",
       "9       0.672035  \n",
       "4       0.668627  \n",
       "28      0.668627  \n",
       "29      0.660846  \n",
       "18      0.643137  \n",
       "0       0.637662  \n",
       "8       0.580420  \n",
       "20      0.568295  \n",
       "10      0.472597  "
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.set_option('display.max_colwidth', None)\n",
    "# pd.reset_option('display.max_colwidth')\n",
    "training_logs[[\"model_name\", \"performance\", \"f1_score_avg\"]].sort_values([\"f1_score_avg\"], ascending=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Top 6 Baseline Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "0        Production-GradientBoostingClassifier\n",
      "26    -oversampling-GradientBoostingClassifier\n",
      "6       -nonbalance-GradientBoostingClassifier\n",
      "17                        -smote-XGBClassifier\n",
      "11               -smote-DecisionTreeClassifier\n",
      "16           -smote-GradientBoostingClassifier\n",
      "27                 -oversampling-XGBClassifier\n",
      "Name: model_name, dtype: object\n",
      "2023-01-07 22:01:08.583887 Based on training logs we can choose 6 top models.\n"
     ]
    }
   ],
   "source": [
    "print(training_logs[[\"model_name\",\"f1_score_avg\"]].sort_values([\"f1_score_avg\"], ascending=False).model_name.iloc[0:7])\n",
    "\n",
    "util.print_debug(message=\"Based on training logs we can choose 6 top models.\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 4. SEARCHING BEST PARAMS FOR ML MODEL"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.1. Gradient Boosting Oversampling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# d_tree = DecisionTreeClassifier(max_depth=2)\n",
    "n_estimators = [1,5,10,30,50,70,100,150,200,250]\n",
    "max_depth = np.arange(2,11,1)\n",
    "learning_rate = np.arange(0.1,1,0.1)\n",
    "\n",
    "param_grid = {\n",
    "        \"n_estimators\": n_estimators,\n",
    "        \"learning_rate\" : learning_rate,\n",
    "        \"max_depth\" : max_depth,\n",
    "        }\n",
    "\n",
    "# Buat gridsearch\n",
    "grad_tree = GradientBoostingClassifier()\n",
    "\n",
    "grad_tree_cv = RandomizedSearchCV(\n",
    "                        estimator = grad_tree,\n",
    "                        param_distributions = param_grid,\n",
    "                        cv = 5,\n",
    "                        scoring = \"precision\",\n",
    "                        n_jobs = -1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "grad_tree_cv.fit(x_train[\"oversampling\"],y_train[\"oversampling\"])\n",
    "\n",
    "grad_tree_cv.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GradientBoostingClassifier(max_depth=7, n_estimators=70, random_state=123)"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grad_tree = GradientBoostingClassifier(\n",
    "                                    max_depth = 7,\n",
    "                                    n_estimators = 70,\n",
    "                                    learning_rate = 0.1,\n",
    "                                    random_state=123)\n",
    "\n",
    "\n",
    "grad_tree.fit(x_train[\"oversampling\"], y_train[\"oversampling\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.93      0.85      0.89        75\n",
      "           1       0.65      0.80      0.71        25\n",
      "\n",
      "    accuracy                           0.84       100\n",
      "   macro avg       0.79      0.83      0.80       100\n",
      "weighted avg       0.86      0.84      0.85       100\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_valid_gb_pred_over = grad_tree.predict(x_valid)\n",
    "\n",
    "print(classification_report(y_true = y_valid,\n",
    "                            y_pred = y_valid_gb_pred_over))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "# Create confusion matrix list\n",
    "disp_cm = list()\n",
    "\n",
    "cm_gb_over = confusion_matrix(y_true = y_valid,\n",
    "                            y_pred = y_valid_gb_pred_over)\n",
    "\n",
    "disp_1 = ConfusionMatrixDisplay(confusion_matrix=cm_gb_over,\n",
    "                               display_labels=grad_tree.classes_)\n",
    "disp_cm.append(disp_1)\n",
    "\n",
    "disp_1.plot()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_valid_gb_pred_over_proba = grad_tree.predict_proba(x_valid)\n",
    "\n",
    "fpr_gb_over, tpr_gb_over, threshold_gb_over = roc_curve(y_valid,\n",
    "                                         y_valid_gb_pred_over_proba[:, 1])\n",
    "roc_auc_gb_over = auc(fpr_gb_over, tpr_gb_over)\n",
    "\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr_gb_over, tpr_gb_over, 'b', label = 'AUC = %0.2f' % roc_auc_gb_over)\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.2. Gradient Boosting Smote"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_estimators = [1,5,10,30,50,70,100,150,200,250]\n",
    "max_depth = np.arange(2,11,1)\n",
    "learning_rate = np.arange(0.1,1,0.1)\n",
    "\n",
    "param_grid = {\n",
    "        \"n_estimators\": n_estimators,\n",
    "        \"learning_rate\" : learning_rate,\n",
    "        \"max_depth\" : max_depth\n",
    "        }\n",
    "\n",
    "\n",
    "grad_tree_cv = RandomizedSearchC(\n",
    "                        estimator = grad_tree,\n",
    "                        param_grid = param_grid,\n",
    "                        cv = 5,\n",
    "                        scoring = \"f1_macro\",\n",
    "                        n_jobs = -1)\n",
    "\n",
    "grid_result = grad_tree_cv.fit(x_train[\"smote\"], y_train[\"smote\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "grid_result.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GradientBoostingClassifier(learning_rate=0.5, max_depth=2, n_estimators=5,\n",
       "                           random_state=123)"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Manual Hyperparameter Tuning\n",
    "grad_tree_smote = GradientBoostingClassifier(\n",
    "                                    max_depth = 2,\n",
    "                                    n_estimators = 5,\n",
    "                                    learning_rate = 0.5,\n",
    "                                    random_state=123)\n",
    "\n",
    "grad_tree_smote.fit(x_train[\"smote\"], y_train[\"smote\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.94      0.80      0.86        75\n",
      "           1       0.58      0.84      0.69        25\n",
      "\n",
      "    accuracy                           0.81       100\n",
      "   macro avg       0.76      0.82      0.78       100\n",
      "weighted avg       0.85      0.81      0.82       100\n",
      "\n"
     ]
    }
   ],
   "source": [
    "## Classification Report\n",
    "y_valid_gb_pred_smote = grad_tree_smote.predict(x_valid)\n",
    "\n",
    "print(classification_report(y_true = y_valid,\n",
    "                            y_pred = y_valid_gb_pred_smote))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAATIAAAEKCAYAAACR79kFAAAAOXRFWHRTb2Z0d2FyZQBNYXRwbG90bGliIHZlcnNpb24zLjUuMSwgaHR0cHM6Ly9tYXRwbG90bGliLm9yZy/YYfK9AAAACXBIWXMAAAsTAAALEwEAmpwYAAAXBUlEQVR4nO3deZQeVZnH8e8vGwESIJ3NBgIBjWJcApzIoo4TlgFUjiCCijhmHBTjIDgqoxnHwV1x5rgCLhHRuAFBRKJgYoyGZQ4CAdkCmCDGJNBkg5iFJUn3M3+81fImdPqtSt6lbvfvc06dt6ret249HQ7PufdW3XsVEZiZpWxAqwMwM9tVTmRmljwnMjNLnhOZmSXPiczMkudEZmbJcyIzs5aRtI+kn0l6SNKDko6W1CZpnqQl2eeIWuU4kZlZK30dmBMRhwCTgAeB6cD8iJgAzM+OeyW/EGtmrSBpL+Ae4OCoSkSS/gRMiYgOSe3Agoh4SW9lDWpsqMWMahsY48cNbnUYVsDiv4xqdQhWwDPPPMnmzZu0K2WceMyesfaJzly/vfPeZxcBz1SdmhERM7L9g4HVwPclTQLuBD4IjI2IDoAsmY2pdZ9SJbLx4wZz+9xxrQ7DCjj+rH9tdQhWwMI7Lt3lMtY80cltc/fP9dvB7X9+JiIm7+DrQcDhwHkRcZukr5OjGdkT95GZWUFBZ3Tl2mpYAayIiNuy459RSWwrsyYl2eeqWgU5kZlZIQF0Ebm2XsuJeBxYLqm7/+s44AFgNjA1OzcVuK5WTKVqWppZGrqoWdvK6zzgJ5KGAI8A76ZSwZol6WxgGXBGrUKcyMyskCDYUrvZmK+siLuBnvrQjitSjhOZmRUSQGeNZmOzOZGZWWG1+r+azYnMzAoJoLNkL9I7kZlZYXXr6q8TJzIzKyQI95GZWdoiYEu58pgTmZkVJTrZpeGadedEZmaFBNDlGpmZpc41MjNLWuWFWCcyM0tYAFuiXPNNOJGZWSGB6CzZxDlOZGZWWFe4aWlmCXMfmZn1AaLTfWRmlrLKDLFOZGaWsAixOQa2OoxtOJGZWWFd7iMzs5RVOvvdtDSzpLmz38wS585+M+sTOv1CrJmlLBBbolypo1zRmFnpubPfzJIXyE1LM0ufO/vNLGkR+PULM0tbpbPfQ5TMLHHu7DezpAXyxIpmlr561cgkLQU2AJ3A1oiYLKkNuAoYDywF3hoRT/ZWTrnqh2ZWepV1LQfk2nI6JiIOjYjJ2fF0YH5ETADmZ8e9ciIzs4IqK43n2XbSKcDMbH8mcGqtC9y0NLNCKsvB5X5qOUrSwqrjGRExY7vifiMpgO9k342NiA6AiOiQNKbWTZzIzKyQCBVpNq6pajL25DUR8ViWrOZJemhnYnIiM7PC6vVCbEQ8ln2uknQtcASwUlJ7VhtrB1bVKsd9ZGZWSGU+MuXaeiNpT0nDu/eBE4D7gdnA1OxnU4HrasXkGpmZFVS3GWLHAtdKgkou+mlEzJF0BzBL0tnAMuCMWgU5kZlZIZXXL3b9hdiIeASY1MP5tcBxRcpyIjOzQjzW0sz6BE/jY2ZJq0zj47GWZpY4Dxo3s6RVZr9w09LMElYZouRE1qdt/NtAvnrBOJY+NBQJPvyVZez/wmf5wrTxrFwxhLH7b+a/vrOU4ft0tjpUAy547y0cedhy1q0fynunvxmAd532R95wzGLWbRgKwOVXHc7t94xrZZgl089qZJJOAr4ODAQui4iLGnm/MvjWhfsxecp6/vu7S9myWTz79ACu/MYYDnvtBt523iquungMV10yhvd8oqPVoRow9+YX8Yt5h/CxaTdvc/6aX0/k6hte0aKoyq/WW/vN1rC0KmkgcCnwemAicKakiY26Xxls2jCA+/6wJye94wkABg8Jhu3dya1z9+b4t1bOHf/WJ7h1zt6tDNOq3PfQC9iwcbdWh5GU7qeWebZmaWSN7Ajg4eztXSRdSWWeoQcaeM+Wevyvu7H3yK18+UMH8MiioUx45dO8/7OP8uSawYwcuxWAkWO3sm6tW/Rld8oJD/FP//BnFj8yim//5FVsfMrJrlrZmpaNjGY/YHnV8Yrs3DYknSNpoaSFq9em3W/U2QkP37cHJ79rDd+ct5ihe3Rx1SU1p1Kykpn920N414fewvs+fgpr1+3OtLPuaHVIpdI9Z3+erVkamch6+ivieSciZkTE5IiYPHpkuYY9FDWqfQuj27dwyOFPAfDak9fx8H27M2LUFtaurNTC1q4cxD4jt7YyTKth3frd6YoBRIgbfv9iXvLC1a0OqVQC2BoDcm3N0sg7rQCqH/XsDzzWwPu1XNuYrYzadzPLH640Q+6+eTgHTHiWo05Yz29ntQHw21ltHH3i31oZptXQts9Tf99/7eRlLF0xooXRlFOd5+zfZY3srLkDmCDpIOBR4O3AOxp4v1I493OP8qUPHMjWLeIFB2zmI19dRnTB56eNZ86VIxmzX+X1CyuHj5+7gEkvfZy9hz/DFRdfxcyfHcakiY/zogPXEiEeXz2Mr13+6laHWS5Nbjbm0bBEFhFbJX0AmEvl9YvLI2JRo+5XFi98+dNcMmfx885/adafWxCN1fKFS6c879ycG1/c/EAS0j2xYpk09PFZRNwA3NDIe5hZ8/WbGpmZ9U31mlixnpzIzKyQQGztKtd7ZE5kZlZYv+ojM7M+KNy0NLPEuY/MzPoEJzIzS1ogOt3Zb2apc2e/mSUt3NlvZn1BOJGZWdr60aBxM+u7XCMzs6RFQGeXE5mZJa5sTy3L9TKImZVeUGla5tnykDRQ0h8l/So7bpM0T9KS7LPmFL1OZGZWUN0XH/kg8GDV8XRgfkRMAOZnx71yIjOzwiLybbVI2h94I3BZ1elTgJnZ/kzg1FrluI/MzAor8NRylKSFVcczImJG1fHXgI8Cw6vOjY2Ijsp9okNSzTUVncjMrJDKU8vcjbk1ETG5py8knQysiog7JU3ZlZicyMyssDzNxhxeA7xJ0huAocBekn4MrJTUntXG2oFVtQpyH5mZFVaPp5YR8Z8RsX9EjKeyXOTvIuKdwGxgavazqcB1teJxjczMCgnyv1qxky4CZkk6G1gGnFHrAicyMyusPi3LqvIiFgALsv21wHFFrnciM7NiAsJDlMwsdR40bmbJq9NTy7rZYSKTdDG9NIUj4vyGRGRmpdY91rJMequRLezlOzPrrwJIJZFFxMzqY0l7RsSmxodkZmVXtqZlzRdiJR0t6QGy0emSJkn6ZsMjM7OSEtGVb2uWPG/2fw04EVgLEBH3AK9rYExmVnaRc2uSXE8tI2K5tE127WxMOGZWepFWZ3+35ZJeDYSkIcD5bDsJmpn1N6n1kQHTgHOB/YBHgUOzYzPrt5Rza46aNbKIWAOc1YRYzCwVXa0OYFt5nloeLOmXklZLWiXpOkkHNyM4Myuh7vfI8mxNkqdp+VNgFtAO7AtcDVzRyKDMrNzqNWd/veRJZIqIH0XE1mz7MaXr6jOzpkrl9QtJbdnu7yVNB66kEtrbgOubEJuZlVVCr1/cSSVxdUf8vqrvAvhso4Iys3JTydpkvY21PKiZgZhZIkKQ4sSKkl4OTKSy0gkAEfHDRgVlZiWXSo2sm6RPAlOoJLIbgNcDtwBOZGb9VckSWZ6nlqdTWQjg8Yh4NzAJ2K2hUZlZuaXy1LLK0xHRJWmrpL2oLJbpF2LN+quUJlasslDSPsB3qTzJ3Ajc3sigzKzcknlq2S0i/i3b/bakOcBeEXFvY8Mys1JLJZFJOry37yLirsaEZGZll1KN7Mu9fBfAsXWOhcX37sGJ+x5a72KtgTa8y899UtJ5T536tlLpI4uIY5oZiJkloslPJPPwAr1mVpwTmZmlTiWbWNGJzMyKK1mNLM8MsZL0TkkXZscHSDqi8aGZWRkp8m+9liMNlXS7pHskLZL06ex8m6R5kpZknyNqxZRniNI3gaOBM7PjDcClOa4zs76qPlNdPwscGxGTqCxqdJKko4DpwPyImADMz457lSeRHRkR5wLPAETEk8CQHNeZWV9Vh7GWUbExOxycbQGcAszMzs8ETq0VTp5EtkXSwO6wJI2mdGuomFkzFWhajpK0sGo7Z5typIGS7qYyhnteRNwGjI2IDoDsc0ytePJ09n8DuBYYI+nzVGbD+ESBv9nM+pIo9NRyTURM3mFREZ3Aodl47muzuQ8LyzPW8ieS7qQylY+AUyPCK42b9Wd1fmoZEeskLQBOAlZKao+IDkntVGprvcrz1PIA4Cngl8BsYFN2zsz6qzr0kUkandXEkLQ7cDzwEJU8MzX72VTgulrh5GlaXs9zi5AMBQ4C/gS8LMe1ZtYH1WnQeDswM+uDHwDMiohfSboVmCXpbGAZcEatgvI0LV9RfZzNivG+HfzczCyXbDqww3o4v5ZKV1Zuhd/sj4i7JL2q6HVm1oeU7M3+PIuPfLjqcABwOLC6YRGZWbkVe2rZFHlqZMOr9rdS6TO7pjHhmFkSUqqRZZ1wwyLiP5oUj5mVnEhohlhJgyJia29TXptZP5VKIqOyUtLhwN2SZgNXA5u6v4yInzc4NjMroxwzWzRbnj6yNmAtlTn6u98nC8CJzKy/Sqizf0z2xPJ+nktg3UqWj82smVKqkQ0EhrFtAutWsj/DzJqqZBmgt0TWERGfaVokZpaGxFZRKtfCdWZWGik1LQuNdTKzfiSVRBYRTzQzEDNLR4pDlMzMnpNYH5mZ2fOI8nWgO5GZWXGukZlZ6lJ6amlm1jMnMjNLWqITK5qZbcs1MjNLnfvIzCx9TmRmljrXyMwsbUFSEyuamT1PUouPmJntkBOZmaVOUa5M5kRmZsV49gsz6wvcR2ZmySvbEKUBrQ7AzBIUObdeSBon6feSHpS0SNIHs/NtkuZJWpJ9jqgVjhOZmRWTrTSeZ6thK/CRiHgpcBRwrqSJwHRgfkRMAOZnx71yIjOz4upQI4uIjoi4K9vfADwI7AecAszMfjYTOLVWOO4jM7NCCr4QO0rSwqrjGREx43llSuOBw4DbgLER0QGVZCdpTK2bOJGZWWHqyp3J1kTE5F7LkoYB1wD/HhHrpeIrArhpaWbF5G1W5sh1kgZTSWI/iYifZ6dXSmrPvm8HVtUqxzWyBhswILh4zmLWdgzmwqkHtzoc286YvTfyqbf+jrbhTxEhfnH7S7nq/17Jsa/4M+89fiHjRz/Juy89jYcerdm66Vfq8fqFKlWv7wEPRsRXqr6aDUwFLso+r6tVVsMSmaTLgZOBVRHx8kbdp+xOfc8ali8Zyh7DOlsdivWgs0t8/fqj+dNjo9ljyGZmnncNty/Zn0ceb+NjPzqR6afd2OoQy6k+L8S+Bvhn4D5Jd2fnPk4lgc2SdDawDDijVkGNrJH9ALgE+GED71Fqo9o3c8Rx67niG2N4yzmrWx2O9WDthj1Zu2FPAJ7aPISlq0cweq9N3P7wuBZHVm71eLM/Im5hx0tkHlekrIb1kUXETcATjSo/BdM+/RiXfa6d6CrbcqbWk/YR63nxvmtYtHxsq0MptwAi8m1N0vLOfknnSFooaeEWnm11OHVz5PHrWbdmEA/ft0erQ7Ecdh+yhYvO+g1f/eWr2fTskFaHU3rqyrc1S8s7+7N3SmYA7KW2kg1F3XkTX7WJo05Yz6uOe4AhuwV7DO/koxf/lf8578BWh2bbGTigk4veOZc5d09gwSI/kKnFEyv2I9//Yjvf/2I7AK88eiOnT1vlJFZKwSdOv5Glq0ZwxS2TWh1MGprcbMzDicz6tUkHPs4bDl/Mko42fnT+1QB8a+4RDB7UxQVvuoV99nyar/7Lr1ncMZIPXn5yi6Mtj35TI5N0BTCFyhCFFcAnI+J7jbpfmd176zDuvXVYq8OwHtzz13aOnD6tx+9uXHRQk6NJSH9JZBFxZqPKNrPW6jc1MjProwLoLFcmcyIzs8JcIzOz9PmppZmlzjUyM0ubl4Mzs9QJkDv7zSx1XmnczNLmpqWZpc9jLc2sD/BTSzNLn2tkZpa08FNLM+sLypXHnMjMrDi/fmFm6XMiM7OkBdDEhUXycCIzs0JEuGlpZn1AV7mqZE5kZlaMm5Zm1he4aWlm6XMiM7O0lW/Q+IBWB2BmieleRSnPVoOkyyWtknR/1bk2SfMkLck+R9Qqx4nMzApTRK4thx8AJ213bjowPyImAPOz4145kZlZcRH5tprFxE3AE9udPgWYme3PBE6tVY77yMysmAC6GtpHNjYiOgAiokPSmFoXOJGZWUGFOvtHSVpYdTwjImbUOyInMjMrLn8iWxMRkwuWvlJSe1YbawdW1brAfWRmVkwAnV35tp0zG5ia7U8Frqt1gWtkZlZQQNRnjJKkK4ApVJqgK4BPAhcBsySdDSwDzqhVjhOZmRVXpxdiI+LMHXx1XJFynMjMrJjGP7UszInMzIor2RAlJzIzK86JzMySFgGdna2OYhtOZGZWnGtkZpY8JzIzS1v4qaWZJS4g6vRCbL04kZlZcTs//KghnMjMrJgILwdnZn2AO/vNLHXhGpmZpa18qyg5kZlZMR40bmapCyA8RMnMkhb1m1ixXpzIzKywcNPSzJJXshqZokRPHyStBv7a6jgaYBSwptVBWCF99b/ZgRExelcKkDSHyr9PHmsiYvuVxOuuVImsr5K0cCeWxLIW8n+ztHg5ODNLnhOZmSXPiaw56r5EvDWc/5slxH1kZpY818jMLHlOZGaWPCeyBpJ0kqQ/SXpY0vRWx2O1Sbpc0ipJ97c6FsvPiaxBJA0ELgVeD0wEzpQ0sbVRWQ4/ABr+AqfVlxNZ4xwBPBwRj0TEZuBK4JQWx2Q1RMRNwBOtjsOKcSJrnP2A5VXHK7JzZlZnTmSNox7O+V0XswZwImucFcC4quP9gcdaFItZn+ZE1jh3ABMkHSRpCPB2YHaLYzLrk5zIGiQitgIfAOYCDwKzImJRa6OyWiRdAdwKvETSCklntzomq81DlMwsea6RmVnynMjMLHlOZGaWPCcyM0ueE5mZJc+JLCGSOiXdLel+SVdL2mMXyvqBpNOz/ct6G9AuaYqkV+/EPZZKet5qOzs6v91vNha816ckXVA0RusbnMjS8nREHBoRLwc2A9Oqv8xm3CgsIt4TEQ/08pMpQOFEZtYsTmTpuhl4UVZb+r2knwL3SRoo6X8l3SHpXknvA1DFJZIekHQ9MKa7IEkLJE3O9k+SdJekeyTNlzSeSsL8UFYb/AdJoyVdk93jDkmvya4dKek3kv4o6Tv0PN50G5J+IelOSYsknbPdd1/OYpkvaXR27oWS5mTX3CzpkLr8a1raIsJbIhuwMfscBFwHvJ9KbWkTcFD23TnAJ7L93YCFwEHAacA8YCCwL7AOOD373QJgMjCayowd3WW1ZZ+fAi6oiuOnwGuz/QOAB7P9bwAXZvtvpDJIflQPf8fS7vNV99gduB8YmR0HcFa2fyFwSbY/H5iQ7R8J/K6nGL31r23QzqU/a5HdJd2d7d8MfI9Kk+/2iPhLdv4E4JXd/V/A3sAE4HXAFRHRCTwm6Xc9lH8UcFN3WRGxo3m5jgcmSn+vcO0laXh2j9Oya6+X9GSOv+l8SW/O9sdlsa4FuoCrsvM/Bn4uaVj2915dde/dctzD+jgnsrQ8HRGHVp/I/ofeVH0KOC8i5m73uzdQexoh5fgNVLokjo6Ip3uIJfeYN0lTqCTFoyPiKUkLgKE7+Hlk9123/b+BmfvI+p65wPslDQaQ9GJJewI3AW/P+tDagWN6uPZW4B8lHZRd25ad3wAMr/rdb6gMiCf73aHZ7k3AWdm51wMjasS6N/BklsQOoVIj7DYA6K5VvgO4JSLWA3+RdEZ2D0maVOMe1g84kfU9lwEPAHdlC2h8h0rN+1pgCXAf8C3gxu0vjIjVVPrYfi7pHp5r2v0SeHN3Zz9wPjA5e5jwAM89Pf008DpJd1Fp4i6rEescYJCke4HPAn+o+m4T8DJJdwLHAp/Jzp8FnJ3FtwhPH2549gsz6wNcIzOz5DmRmVnynMjMLHlOZGaWPCcyM0ueE5mZJc+JzMyS9//WBtR2y/PQVgAAAABJRU5ErkJggg==",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cm_gb_smote = confusion_matrix(y_true = y_valid,\n",
    "                            y_pred = y_valid_gb_pred_smote)\n",
    "\n",
    "disp_2 = ConfusionMatrixDisplay(confusion_matrix=cm_gb_smote,\n",
    "                               display_labels=grad_tree.classes_)\n",
    "\n",
    "disp_cm.append(disp_2)\n",
    "\n",
    "disp_2.plot()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": true
   },
   "outputs": [
    {
     "ename": "NameError",
     "evalue": "name 'grid_result' is not defined",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mNameError\u001b[0m                                 Traceback (most recent call last)",
      "Input \u001b[1;32mIn [28]\u001b[0m, in \u001b[0;36m<cell line: 1>\u001b[1;34m()\u001b[0m\n\u001b[1;32m----> 1\u001b[0m means \u001b[38;5;241m=\u001b[39m \u001b[43mgrid_result\u001b[49m\u001b[38;5;241m.\u001b[39mcv_results_[\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mmean_test_score\u001b[39m\u001b[38;5;124m'\u001b[39m]\n\u001b[0;32m      2\u001b[0m stds \u001b[38;5;241m=\u001b[39m grid_result\u001b[38;5;241m.\u001b[39mcv_results_[\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mstd_test_score\u001b[39m\u001b[38;5;124m'\u001b[39m]\n\u001b[0;32m      3\u001b[0m params \u001b[38;5;241m=\u001b[39m grid_result\u001b[38;5;241m.\u001b[39mcv_results_[\u001b[38;5;124m'\u001b[39m\u001b[38;5;124mparams\u001b[39m\u001b[38;5;124m'\u001b[39m]\n",
      "\u001b[1;31mNameError\u001b[0m: name 'grid_result' is not defined"
     ]
    }
   ],
   "source": [
    "# means = grid_result.cv_results_['mean_test_score']\n",
    "# stds = grid_result.cv_results_['std_test_score']\n",
    "# params = grid_result.cv_results_['params']\n",
    "\n",
    "# print(\"Best: %f using %s\" % (grid_result.best_score_, grid_result.best_params_))\n",
    "\n",
    "# # plot results\n",
    "# scores = numpy.array(means).reshape(2, 5)\n",
    "# plt.subplots(figsize=(20,15))\n",
    "# for i, value in enumerate(max_depth):\n",
    "#     plt.plot(n_estimators, scores[i], label='depth: ' + str(value))\n",
    "# plt.legend()\n",
    "# plt.xlabel('n_estimators')\n",
    "# plt.ylabel('Log Loss')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.3 ROC Curve Gradient Boosting on SMOTE and Oversampling dataset"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "y_valid_gb_pred_over_proba = grad_tree.predict_proba(x_valid)\n",
    "y_valid_gb_pred_smote_proba = grad_tree_smote.predict_proba(x_valid)\n",
    "\n",
    "fpr_gb_over, tpr_gb_over, threshold_gb_over = roc_curve(y_valid,\n",
    "                                         y_valid_gb_pred_over_proba[:, 1])\n",
    "\n",
    "fpr_gb_smote, tpr_gb_smote, threshold_gb_smote = roc_curve(y_valid,\n",
    "                                         y_valid_gb_pred_smote_proba[:, 1])\n",
    "\n",
    "roc_auc_gb_over = auc(fpr_gb_over, tpr_gb_over)\n",
    "roc_auc_gb_smote = auc(fpr_gb_smote, tpr_gb_smote)\n",
    "\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "plt.plot(fpr_gb_over, tpr_gb_over, 'b', label = 'AUC GB Over = %0.2f' % roc_auc_gb_over)\n",
    "plt.plot(fpr_gb_smote, tpr_gb_smote, 'g', label = 'AUC GB Smote = %0.2f' % roc_auc_gb_smote)\n",
    "\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# util.pickle_dump(grad_tree, config_data['model_params_path'][4])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.4. XGBOOST Smote"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "n_estimators = [1,10,30,50,70,100,150,200,250,300]\n",
    "max_depth = np.arange(2,11,1)\n",
    "learning_rate = np.arange(0.1,1,0.1)\n",
    "scale_pos_weight = np.arange(0.1,0.5,0.1)\n",
    "gamma = np.arange(1,6,1)\n",
    "\n",
    "xgb = XGBClassifier()\n",
    "\n",
    "dist_params_xgb1 = {\n",
    "        \"n_estimators\" : n_estimators,\n",
    "        \"learning_rate\" : learning_rate,\n",
    "        \"max_depth\" : max_depth,\n",
    "        \"scale_pos_weight\" : scale_pos_weight,\n",
    "        \"gamma\" : gamma,\n",
    "        \"random_state\" : [123]\n",
    "    }\n",
    "\n",
    "xgb_cv = RandomizedSearchCV(estimator=xgb,\n",
    "                      param_distributions = dist_params_xgb1,\n",
    "                      cv = 5,\n",
    "                      scoring = 'recall',\n",
    "                      n_jobs = -1,\n",
    "                      refit = True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "xgb_cv.fit(x_train[\"smote\"], y_train[\"smote\"])\n",
    "\n",
    "xgb_cv.best_params_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "XGBClassifier(alpha=5, base_score=0.5, booster='gbtree', callbacks=None,\n",
       "              colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,\n",
       "              early_stopping_rounds=None, enable_categorical=False,\n",
       "              eval_metric=None, feature_types=None, gamma=0, gpu_id=-1,\n",
       "              grow_policy='depthwise', importance_type=None,\n",
       "              interaction_constraints='', learning_rate=0.7, max_bin=256,\n",
       "              max_cat_threshold=64, max_cat_to_onehot=4, max_delta_step=0,\n",
       "              max_depth=4, max_leaves=0, min_child_weight=1, missing=nan,\n",
       "              monotone_constraints='()', n_estimators=7, n_jobs=0,\n",
       "              num_parallel_tree=1, predictor='auto', ...)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xgb_smote = XGBClassifier(\n",
    "                    n_estimators = 7,\n",
    "                    learning_rate = 0.7,\n",
    "                    max_depth = 4,\n",
    "                    scale_pos_weight = 0.4,\n",
    "                    alpha = 5,\n",
    "                    random_state=123)\n",
    "\n",
    "xgb_smote.fit(x_train[\"smote\"], y_train[\"smote\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.87      0.91      0.89        75\n",
      "           1       0.68      0.60      0.64        25\n",
      "\n",
      "    accuracy                           0.83       100\n",
      "   macro avg       0.78      0.75      0.76       100\n",
      "weighted avg       0.82      0.83      0.83       100\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_valid_xgb_pred_smote = xgb_smote.predict(x_valid)\n",
    "\n",
    "print(classification_report(y_true = y_valid,\n",
    "                            y_pred = y_valid_xgb_pred_smote))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "ename": "AttributeError",
     "evalue": "'numpy.ndarray' object has no attribute 'classes_'",
     "output_type": "error",
     "traceback": [
      "\u001b[1;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[1;31mAttributeError\u001b[0m                            Traceback (most recent call last)",
      "Input \u001b[1;32mIn [10]\u001b[0m, in \u001b[0;36m<cell line: 4>\u001b[1;34m()\u001b[0m\n\u001b[0;32m      1\u001b[0m cm_xgb_smote \u001b[38;5;241m=\u001b[39m confusion_matrix(y_true \u001b[38;5;241m=\u001b[39m y_valid,\n\u001b[0;32m      2\u001b[0m                             y_pred \u001b[38;5;241m=\u001b[39m y_valid_xgb_pred_smote)\n\u001b[0;32m      4\u001b[0m disp_3 \u001b[38;5;241m=\u001b[39m ConfusionMatrixDisplay(confusion_matrix\u001b[38;5;241m=\u001b[39mcm_xgb_smote,\n\u001b[1;32m----> 5\u001b[0m                                display_labels\u001b[38;5;241m=\u001b[39m\u001b[43mcm_xgb_smote\u001b[49m\u001b[38;5;241;43m.\u001b[39;49m\u001b[43mclasses_\u001b[49m)\n\u001b[0;32m      6\u001b[0m disp_cm\u001b[38;5;241m.\u001b[39mappend(disp_3)\n\u001b[0;32m      8\u001b[0m disp_3\u001b[38;5;241m.\u001b[39mplot()\n",
      "\u001b[1;31mAttributeError\u001b[0m: 'numpy.ndarray' object has no attribute 'classes_'"
     ]
    }
   ],
   "source": [
    "cm_xgb_smote = confusion_matrix(y_true = y_valid,\n",
    "                            y_pred = y_valid_xgb_pred_smote)\n",
    "\n",
    "disp_3 = ConfusionMatrixDisplay(confusion_matrix=cm_xgb_smote,\n",
    "                               display_labels=grad_tree.classes_)\n",
    "disp_cm.append(disp_3)\n",
    "\n",
    "disp_3.plot()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import seaborn as sns\n",
    "\n",
    "train_set = pd.concat([x_train['nonbalance'], y_train['nonbalance']], axis=1)\n",
    "\n",
    "ax = sns.histplot(data=train_set, x='number_of_vehicles_involved', multiple='stack', hue='fraud_reported')\n",
    "ax.set(ylabel='number of fraud')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 1800x432 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig, ax = plt.subplots(figsize=(20,6))\n",
    "\n",
    "feat_imp = pd.Series(xgb_smote.feature_importances_, x_valid.columns.to_list()).sort_values(ascending=False)\n",
    "feat_imp.plot(kind='bar', title='Feature Importances')\n",
    "plt.ylabel('Feature Importance Score')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.5. XGBoosting Oversampling"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "XGBClassifier(alpha=5, base_score=0.5, booster='gbtree', callbacks=None,\n",
       "              colsample_bylevel=1, colsample_bynode=1, colsample_bytree=1,\n",
       "              early_stopping_rounds=None, enable_categorical=False,\n",
       "              eval_metric=None, feature_types=None, gamma=0, gpu_id=-1,\n",
       "              grow_policy='depthwise', importance_type=None,\n",
       "              interaction_constraints='', learning_rate=0.7, max_bin=256,\n",
       "              max_cat_threshold=64, max_cat_to_onehot=4, max_delta_step=0,\n",
       "              max_depth=4, max_leaves=0, min_child_weight=1, missing=nan,\n",
       "              monotone_constraints='()', n_estimators=7, n_jobs=0,\n",
       "              num_parallel_tree=1, predictor='auto', ...)"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "xgb_over = XGBClassifier(\n",
    "                    n_estimators = 7,\n",
    "                    learning_rate = 0.7,\n",
    "                    max_depth = 4,\n",
    "                    scale_pos_weight = 0.4,\n",
    "                    alpha = 5,\n",
    "                    random_state=123)\n",
    "\n",
    "xgb_over.fit(x_train[\"oversampling\"], y_train[\"oversampling\"])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.85      0.85      0.85        75\n",
      "           1       0.56      0.56      0.56        25\n",
      "\n",
      "    accuracy                           0.78       100\n",
      "   macro avg       0.71      0.71      0.71       100\n",
      "weighted avg       0.78      0.78      0.78       100\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_valid_xgb_pred_over = xgb_over.predict(x_valid)\n",
    "\n",
    "print(classification_report(y_true = y_valid,\n",
    "                            y_pred = y_valid_xgb_pred_over))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cm_xgb_over = confusion_matrix(y_true = y_valid,\n",
    "                            y_pred = y_valid_xgb_pred_over)\n",
    "\n",
    "disp_4 = ConfusionMatrixDisplay(confusion_matrix=cm_xgb_over,\n",
    "                               display_labels=grad_tree.classes_)\n",
    "\n",
    "disp_cm.append(disp_4)\n",
    "disp_4.plot()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# util.pickle_dump(xgb, config_data['model_params_path'][6])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.6. Decision Tree"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9194352159468439"
      ]
     },
     "execution_count": 37,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dt_tree = DecisionTreeClassifier(\n",
    "                                max_depth = 5,\n",
    "                                random_state = 123)\n",
    "\n",
    "dt_tree.fit(x_train[\"smote\"], y_train[\"smote\"])\n",
    "\n",
    "y_dt_pred = dt_tree.predict(x_train[\"smote\"])\n",
    "\n",
    "accuracy_score(y_true=y_train[\"smote\"],\n",
    "               y_pred=y_dt_pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.95      0.79      0.86        75\n",
      "           1       0.58      0.88      0.70        25\n",
      "\n",
      "    accuracy                           0.81       100\n",
      "   macro avg       0.77      0.83      0.78       100\n",
      "weighted avg       0.86      0.81      0.82       100\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_valid_dt_smote = dt_tree.predict(x_valid)\n",
    "\n",
    "print(classification_report(y_true = y_valid,\n",
    "                            y_pred = y_valid_dt_smote))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cm_dt = confusion_matrix(y_true = y_valid,\n",
    "                            y_pred = y_valid_dt_smote)\n",
    "\n",
    "disp_5 = ConfusionMatrixDisplay(confusion_matrix=cm_dt,\n",
    "                               display_labels=grad_tree.classes_)\n",
    "\n",
    "disp_cm.append(disp_5)\n",
    "disp_5.plot()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 4.7. ADABOOST"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "metadata": {},
   "outputs": [],
   "source": [
    "    dist_params_ada = {\n",
    "        \"base_estimator\" : [DecisionTreeClassifier(max_depth = 2)],\n",
    "        \"n_estimators\" : [3],\n",
    "        \"learning_rate\" : [0.7],\n",
    "        \"random_state\": [123],\n",
    "    }"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "AdaBoostClassifier(base_estimator=DecisionTreeClassifier(max_depth=2),\n",
       "                   learning_rate=0.7, n_estimators=3, random_state=123)"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ada_over = AdaBoostClassifier(base_estimator = DecisionTreeClassifier(max_depth = 2),\n",
    "                             n_estimators = 3,\n",
    "                             learning_rate = 0.7,\n",
    "                             random_state = 123)\n",
    "\n",
    "ada_over.fit(x_train['oversampling'], y_train['oversampling'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       0.97      0.79      0.87        75\n",
      "           1       0.59      0.92      0.72        25\n",
      "\n",
      "    accuracy                           0.82       100\n",
      "   macro avg       0.78      0.85      0.79       100\n",
      "weighted avg       0.87      0.82      0.83       100\n",
      "\n"
     ]
    }
   ],
   "source": [
    "y_valid_ada_over = ada_over.predict(x_valid)\n",
    "\n",
    "print(classification_report(y_true = y_valid,\n",
    "                            y_pred = y_valid_ada_over))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 432x288 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "cm_ada = confusion_matrix(y_true = y_valid,\n",
    "                      y_pred = y_valid_ada_over)\n",
    "\n",
    "disp_6 = ConfusionMatrixDisplay(confusion_matrix=cm_ada,\n",
    "                               display_labels=ada_over.classes_)\n",
    "\n",
    "disp_cm.append(disp_6)\n",
    "disp_6.plot()\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 5. ROC AUC PLOT"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 69,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "## Create Proba for all function\n",
    "y_valid_xgb_over_proba = xgb_over.predict_proba(x_valid)\n",
    "y_valid_xgb_smote_proba = xgb_smote.predict_proba(x_valid)\n",
    "y_valid_dt_proba = dt_tree.predict_proba(x_valid)\n",
    "y_valid_ada_over_proba = ada_over.predict_proba(x_valid)\n",
    "y_valid_gb_pred_over_proba = grad_tree.predict_proba(x_valid)\n",
    "y_valid_gb_pred_smote_proba = grad_tree_smote.predict_proba(x_valid)\n",
    "\n",
    "## ROC CURVE Function for create FPR & TPR\n",
    "fpr_xgb_over, tpr_xgb_over, threshold_xgb_over = roc_curve(y_valid,\n",
    "                                         y_valid_xgb_over_proba[:, 1])\n",
    "fpr_xgb_smote, tpr_xgb_smote, threshold_xgb_smote = roc_curve(y_valid,\n",
    "                                         y_valid_xgb_smote_proba[:, 1])\n",
    "fpr_dt, tpr_dt, threshold_dt = roc_curve(y_valid,\n",
    "                                        y_valid_dt_proba[:, 1])\n",
    "fpr_ada_over, tpr_ada_over, threshold_ada_over = roc_curve(y_valid,\n",
    "                                                           y_valid_ada_over_proba[:, 1])\n",
    "fpr_gb_over, tpr_gb_over, threshold_gb_over = roc_curve(y_valid,\n",
    "                                         y_valid_gb_pred_over_proba[:, 1])\n",
    "\n",
    "fpr_gb_smote, tpr_gb_smote, threshold_gb_smote = roc_curve(y_valid,\n",
    "                                         y_valid_gb_pred_smote_proba[:, 1])\n",
    "\n",
    "roc_auc_xgb_over = auc(fpr_xgb_over, tpr_xgb_over)\n",
    "roc_auc_xgb_smote = auc(fpr_xgb_smote, tpr_xgb_smote)\n",
    "roc_auc_dt = auc(fpr_dt, tpr_dt)\n",
    "roc_auc_ada_over = auc(fpr_ada_over, tpr_ada_over)\n",
    "roc_auc_gb_over = auc(fpr_gb_over, tpr_gb_over)\n",
    "roc_auc_gb_smote = auc(fpr_gb_smote, tpr_gb_smote)\n",
    "\n",
    "\n",
    "plt.subplots(figsize=(10,8))\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "\n",
    "plt.plot(fpr_gb_over, tpr_gb_over, 'b', label = 'AUC GradientBoost Oversampling = %0.2f' % roc_auc_gb_over)\n",
    "plt.plot(fpr_gb_smote, tpr_gb_smote, 'g', label = 'AUC GradientBoost Smote = %0.2f' % roc_auc_gb_smote)\n",
    "plt.plot(fpr_xgb_over, tpr_xgb_over, 'k', label = 'AUC XGBoost Oversampling = %0.2f' % roc_auc_xgb_over)\n",
    "plt.plot(fpr_xgb_smote, tpr_xgb_smote, 'r', label = 'AUC XGBoost Smote = %0.2f' % roc_auc_xgb_smote)\n",
    "plt.plot(fpr_dt, tpr_dt, 'y', label = 'AUC DecisionTree = %0.2f' % roc_auc_dt)\n",
    "plt.plot(fpr_ada_over, tpr_ada_over, 'c', label = 'AUC AdaBoost smote = %0.2f' % roc_auc_ada_over)\n",
    "\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 5.1. Confusion Matrix"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "# fig, axes = plt.subplots(ncols=3, nrows=3, figsize=(15,12), sharex=False, sharey=False)\n",
    "# axes = axes.ravel()\n",
    "# cols = \n",
    "labels = ['gb_over', 'gb_smote', 'xgb_smote', 'xgb_over', 'dt', 'ada']\n",
    "\n",
    "# disp_cm_test = disp_cm.copy()\n",
    "\n",
    "for cm, label in zip(disp_cm, labels):\n",
    "    plt.title(label)\n",
    "    cm.plot()\n",
    "\n",
    "    plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 6. Hyperparameter Tuning"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6.1 Create Param Distribution"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 71,
   "metadata": {},
   "outputs": [],
   "source": [
    "def create_dist_params(model_name: str) -> dict:\n",
    "    # Define models paramteres   \n",
    "    dist_params_xgb = {\n",
    "        \"n_estimators\" : [7],\n",
    "        \"learning_rate\" : [0.7],\n",
    "        \"max_depth\" : [4],\n",
    "        \"alpha\" : [5],\n",
    "        \"scale_pos_weight\" : [0.4],\n",
    "    }\n",
    "    dist_params_gb = {\n",
    "        \"max_depth\" : [2],\n",
    "        \"n_estimators\" : [5],\n",
    "        \"learning_rate\" : [0.5],\n",
    "    }\n",
    "    dist_params_ada = {\n",
    "        \"base_estimator\" : [DecisionTreeClassifier(max_depth = 2)],\n",
    "        \"n_estimators\" : [3],\n",
    "        \"learning_rate\" : [0.7],\n",
    "    }\n",
    "    dist_params_dt = {\n",
    "        \"max_depth\" : [5],\n",
    "    }\n",
    "\n",
    "    # Make all models parameters in to one\n",
    "    dist_params = {\n",
    "        \"XGBClassifier\": dist_params_xgb,\n",
    "        \"GradientBoostingClassifier\": dist_params_gb,\n",
    "        \"AdaBoostClassifier\": dist_params_ada,\n",
    "        \"DecisionTreeClassifier\": dist_params_dt\n",
    "    }\n",
    "\n",
    "    # Return distribution of model parameters\n",
    "    return dist_params[model_name]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 72,
   "metadata": {},
   "outputs": [],
   "source": [
    "def hyper_params_tuning(model: list) -> list:\n",
    "    # Create copy of current best baseline model\n",
    "    model_list = []\n",
    "    trained_model = [XGBClassifier(),GradientBoostingClassifier(),AdaBoostClassifier(),DecisionTreeClassifier()]\n",
    "    \n",
    "    # Create model object\n",
    "    for col, mod in list(zip(model, trained_model)):\n",
    "        dist_params = create_dist_params(col)\n",
    "        model_rsc = GridSearchCV(mod, dist_params, n_jobs = -1)\n",
    "        model_ = {\n",
    "                        \"model_name\": col,\n",
    "                        \"model_object\": model_rsc,\n",
    "                        \"model_uid\": \"\"\n",
    "                    }\n",
    "            \n",
    "        model_list.append(model_.copy())\n",
    "    \n",
    "    # Return model object\n",
    "    return model_list"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.2 run debugging"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {
    "scrolled": true
   },
   "outputs": [],
   "source": [
    "listmodel_ = hyper_params_tuning(['XGBClassifier',\n",
    "                                  'GradientBoostingClassifier', \n",
    "                                  'AdaBoostClassifier',\n",
    "                                  'DecisionTreeClassifier'])\n",
    "\n",
    "listmodel_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "## Sanity Check\n",
    "x_test.shape == x_valid.shape"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6.3 Create Training Function based on Validation"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_eval_valid(configuration_model: str, params: dict, hyperparams_model: list = None):\n",
    "    # Load dataset\n",
    "    # x_train, y_train, \\\n",
    "    # x_valid, y_valid, \\\n",
    "    # x_test, y_test = load_dataset(params)\n",
    "\n",
    "    # Variabel to store trained models\n",
    "    list_of_trained_model = dict()\n",
    "\n",
    "    # Create log template\n",
    "    training_log = training_log_template()\n",
    "\n",
    "    # Training for every data configuration\n",
    "    for config_data in x_train:\n",
    "        # Debug message\n",
    "        util.print_debug(\"Training model based on configuration data: {}\".format(config_data))\n",
    "\n",
    "        # Create model objects\n",
    "        if hyperparams_model == None:\n",
    "            list_of_model = create_model_object(params)\n",
    "        else:\n",
    "            list_of_model = copy.deepcopy(hyperparams_model)\n",
    "\n",
    "        # Variabel to store tained model\n",
    "        trained_model = list()\n",
    "\n",
    "        # Load train data based on its configuration\n",
    "        x_train_data = x_train[config_data]\n",
    "        y_train_data = y_train[config_data]\n",
    "\n",
    "        # Train each model by current dataset configuration\n",
    "        for model in list_of_model:\n",
    "            # Debug message\n",
    "            util.print_debug(\"Training model: {}\".format(model[\"model_name\"]))\n",
    "\n",
    "            # Training\n",
    "            training_time = util.time_stamp()\n",
    "            model[\"model_object\"].fit(x_train_data, y_train_data)\n",
    "            training_time = (util.time_stamp() - training_time).total_seconds()\n",
    "\n",
    "            # Debug message\n",
    "            util.print_debug(\"Evalutaing model: {}\".format(model[\"model_name\"]))\n",
    "\n",
    "            # Evaluation\n",
    "            y_predict = model[\"model_object\"].predict(x_valid)\n",
    "            performance = classification_report(y_valid, y_predict, output_dict = True)\n",
    "\n",
    "            # Debug message\n",
    "            util.print_debug(\"Logging: {}\".format(model[\"model_name\"]))\n",
    "\n",
    "            # Create UID\n",
    "            uid = hashlib.md5(str(training_time).encode()).hexdigest()\n",
    "\n",
    "            # Assign model's UID\n",
    "            model[\"model_uid\"] = uid\n",
    "\n",
    "            # Create training log data\n",
    "            training_log[\"model_name\"].append(\"{}-{}-{}\".format(configuration_model, config_data, model[\"model_name\"]))\n",
    "            training_log[\"model_uid\"].append(uid)\n",
    "            training_log[\"training_time\"].append(training_time)\n",
    "            training_log[\"training_date\"].append(util.time_stamp())\n",
    "            training_log[\"performance\"].append(performance)\n",
    "            training_log[\"f1_score_avg\"].append(performance[\"weighted avg\"][\"recall\"])\n",
    "            training_log[\"data_configurations\"].append(config_data)\n",
    "\n",
    "            # Collect current trained model\n",
    "            trained_model.append(copy.deepcopy(model))\n",
    "\n",
    "            # Debug message\n",
    "            util.print_debug(\"Model {} has been trained for configuration data {}.\".format(model[\"model_name\"], config_data))\n",
    "        \n",
    "        # Collect current trained list of model\n",
    "        list_of_trained_model[config_data] = copy.deepcopy(trained_model)\n",
    "    \n",
    "    \n",
    "    # Debug message\n",
    "    util.print_debug(\"All combination models and configuration data has been trained.\")\n",
    "    \n",
    "    # Return list trained model\n",
    "    return list_of_trained_model, training_log"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 6.4 Training model and update Training Logs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 73,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2023-01-07 22:15:31.071627 creating training log template.\n",
      "2023-01-07 22:15:31.071627 Training log template created.\n",
      "2023-01-07 22:15:31.071627 Training model based on configuration data: nonbalance\n",
      "2023-01-07 22:15:31.073625 Training model: XGBClassifier\n",
      "2023-01-07 22:15:31.917829 Evalutaing model: XGBClassifier\n",
      "2023-01-07 22:15:31.964822 Logging: XGBClassifier\n",
      "2023-01-07 22:15:31.974823 Model XGBClassifier has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:15:31.974823 Training model: GradientBoostingClassifier\n",
      "2023-01-07 22:15:32.217328 Evalutaing model: GradientBoostingClassifier\n",
      "2023-01-07 22:15:32.237357 Logging: GradientBoostingClassifier\n",
      "2023-01-07 22:15:32.244982 Model GradientBoostingClassifier has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:15:32.244982 Training model: AdaBoostClassifier\n",
      "2023-01-07 22:15:32.498412 Evalutaing model: AdaBoostClassifier\n",
      "2023-01-07 22:15:32.547412 Logging: AdaBoostClassifier\n",
      "2023-01-07 22:15:32.553410 Model AdaBoostClassifier has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:15:32.554412 Training model: DecisionTreeClassifier\n",
      "2023-01-07 22:15:32.725409 Evalutaing model: DecisionTreeClassifier\n",
      "2023-01-07 22:15:32.755412 Logging: DecisionTreeClassifier\n",
      "2023-01-07 22:15:32.759408 Model DecisionTreeClassifier has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:15:32.779409 Training model based on configuration data: smote\n",
      "2023-01-07 22:15:32.782410 Training model: XGBClassifier\n",
      "2023-01-07 22:15:33.735446 Evalutaing model: XGBClassifier\n",
      "2023-01-07 22:15:33.774447 Logging: XGBClassifier\n",
      "2023-01-07 22:15:33.787448 Model XGBClassifier has been trained for configuration data smote.\n",
      "2023-01-07 22:15:33.788446 Training model: GradientBoostingClassifier\n",
      "2023-01-07 22:15:34.422448 Evalutaing model: GradientBoostingClassifier\n",
      "2023-01-07 22:15:34.451450 Logging: GradientBoostingClassifier\n",
      "2023-01-07 22:15:34.458447 Model GradientBoostingClassifier has been trained for configuration data smote.\n",
      "2023-01-07 22:15:34.458447 Training model: AdaBoostClassifier\n",
      "2023-01-07 22:15:35.027451 Evalutaing model: AdaBoostClassifier\n",
      "2023-01-07 22:15:35.054447 Logging: AdaBoostClassifier\n",
      "2023-01-07 22:15:35.058447 Model AdaBoostClassifier has been trained for configuration data smote.\n",
      "2023-01-07 22:15:35.059449 Training model: DecisionTreeClassifier\n",
      "2023-01-07 22:15:35.366225 Evalutaing model: DecisionTreeClassifier\n",
      "2023-01-07 22:15:35.389951 Logging: DecisionTreeClassifier\n",
      "2023-01-07 22:15:35.392951 Model DecisionTreeClassifier has been trained for configuration data smote.\n",
      "2023-01-07 22:15:35.411951 Training model based on configuration data: oversampling\n",
      "2023-01-07 22:15:35.414951 Training model: XGBClassifier\n",
      "2023-01-07 22:15:36.237386 Evalutaing model: XGBClassifier\n",
      "2023-01-07 22:15:36.275386 Logging: XGBClassifier\n",
      "2023-01-07 22:15:36.282386 Model XGBClassifier has been trained for configuration data oversampling.\n",
      "2023-01-07 22:15:36.282386 Training model: GradientBoostingClassifier\n",
      "2023-01-07 22:15:36.728383 Evalutaing model: GradientBoostingClassifier\n",
      "2023-01-07 22:15:36.753812 Logging: GradientBoostingClassifier\n",
      "2023-01-07 22:15:36.759814 Model GradientBoostingClassifier has been trained for configuration data oversampling.\n",
      "2023-01-07 22:15:36.760813 Training model: AdaBoostClassifier\n",
      "2023-01-07 22:15:37.268518 Evalutaing model: AdaBoostClassifier\n",
      "2023-01-07 22:15:37.304518 Logging: AdaBoostClassifier\n",
      "2023-01-07 22:15:37.310520 Model AdaBoostClassifier has been trained for configuration data oversampling.\n",
      "2023-01-07 22:15:37.310520 Training model: DecisionTreeClassifier\n",
      "2023-01-07 22:15:37.756579 Evalutaing model: DecisionTreeClassifier\n",
      "2023-01-07 22:15:37.791583 Logging: DecisionTreeClassifier\n",
      "2023-01-07 22:15:37.795580 Model DecisionTreeClassifier has been trained for configuration data oversampling.\n",
      "2023-01-07 22:15:37.878577 All combination models and configuration data has been trained.\n"
     ]
    }
   ],
   "source": [
    "list_of_trained_model_valid, \\\n",
    "training_log_valid = train_eval_valid(\"Hyperparams\", \n",
    "                                    config_data,\n",
    "                                    hyper_params_tuning(['XGBClassifier',\n",
    "                                                          'GradientBoostingClassifier', \n",
    "                                                          'AdaBoostClassifier',\n",
    "                                                          'DecisionTreeClassifier']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 6.4.1. Display Best Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model_name</th>\n",
       "      <th>training_time</th>\n",
       "      <th>performance</th>\n",
       "      <th>f1_score_avg</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Hyperparams-smote-XGBClassifier</td>\n",
       "      <td>0.952034</td>\n",
       "      <td>{'0': {'precision': 0.8717948717948718, 'recall': 0.9066666666666666, 'f1-score': 0.8888888888888887, 'support': 75}, '1': {'precision': 0.6818181818181818, 'recall': 0.6, 'f1-score': 0.6382978723404256, 'support': 25}, 'accuracy': 0.83, 'macro avg': {'precision': 0.7768065268065267, 'recall': 0.7533333333333333, 'f1-score': 0.7635933806146571, 'support': 100}, 'weighted avg': {'precision': 0.8243006993006994, 'recall': 0.83, 'f1-score': 0.826241134751773, 'support': 100}}</td>\n",
       "      <td>0.83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>Hyperparams-oversampling-GradientBoostingClassifier</td>\n",
       "      <td>0.445997</td>\n",
       "      <td>{'0': {'precision': 0.9672131147540983, 'recall': 0.7866666666666666, 'f1-score': 0.8676470588235293, 'support': 75}, '1': {'precision': 0.5897435897435898, 'recall': 0.92, 'f1-score': 0.71875, 'support': 25}, 'accuracy': 0.82, 'macro avg': {'precision': 0.7784783522488441, 'recall': 0.8533333333333333, 'f1-score': 0.7931985294117647, 'support': 100}, 'weighted avg': {'precision': 0.8728457335014713, 'recall': 0.82, 'f1-score': 0.830422794117647, 'support': 100}}</td>\n",
       "      <td>0.82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Hyperparams-oversampling-AdaBoostClassifier</td>\n",
       "      <td>0.507705</td>\n",
       "      <td>{'0': {'precision': 0.9672131147540983, 'recall': 0.7866666666666666, 'f1-score': 0.8676470588235293, 'support': 75}, '1': {'precision': 0.5897435897435898, 'recall': 0.92, 'f1-score': 0.71875, 'support': 25}, 'accuracy': 0.82, 'macro avg': {'precision': 0.7784783522488441, 'recall': 0.8533333333333333, 'f1-score': 0.7931985294117647, 'support': 100}, 'weighted avg': {'precision': 0.8728457335014713, 'recall': 0.82, 'f1-score': 0.830422794117647, 'support': 100}}</td>\n",
       "      <td>0.82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Hyperparams-smote-AdaBoostClassifier</td>\n",
       "      <td>0.569004</td>\n",
       "      <td>{'0': {'precision': 0.9384615384615385, 'recall': 0.8133333333333334, 'f1-score': 0.8714285714285714, 'support': 75}, '1': {'precision': 0.6, 'recall': 0.84, 'f1-score': 0.7000000000000001, 'support': 25}, 'accuracy': 0.82, 'macro avg': {'precision': 0.7692307692307692, 'recall': 0.8266666666666667, 'f1-score': 0.7857142857142858, 'support': 100}, 'weighted avg': {'precision': 0.8538461538461539, 'recall': 0.82, 'f1-score': 0.8285714285714286, 'support': 100}}</td>\n",
       "      <td>0.82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Hyperparams-nonbalance-AdaBoostClassifier</td>\n",
       "      <td>0.252431</td>\n",
       "      <td>{'0': {'precision': 0.9242424242424242, 'recall': 0.8133333333333334, 'f1-score': 0.8652482269503545, 'support': 75}, '1': {'precision': 0.5882352941176471, 'recall': 0.8, 'f1-score': 0.6779661016949153, 'support': 25}, 'accuracy': 0.81, 'macro avg': {'precision': 0.7562388591800356, 'recall': 0.8066666666666666, 'f1-score': 0.771607164322635, 'support': 100}, 'weighted avg': {'precision': 0.8402406417112299, 'recall': 0.81, 'f1-score': 0.8184276956364946, 'support': 100}}</td>\n",
       "      <td>0.81</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Hyperparams-smote-GradientBoostingClassifier</td>\n",
       "      <td>0.633001</td>\n",
       "      <td>{'0': {'precision': 0.9375, 'recall': 0.8, 'f1-score': 0.8633093525179856, 'support': 75}, '1': {'precision': 0.5833333333333334, 'recall': 0.84, 'f1-score': 0.6885245901639344, 'support': 25}, 'accuracy': 0.81, 'macro avg': {'precision': 0.7604166666666667, 'recall': 0.8200000000000001, 'f1-score': 0.7759169713409599, 'support': 100}, 'weighted avg': {'precision': 0.8489583333333333, 'recall': 0.81, 'f1-score': 0.8196131619294728, 'support': 100}}</td>\n",
       "      <td>0.81</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Hyperparams-nonbalance-DecisionTreeClassifier</td>\n",
       "      <td>0.170997</td>\n",
       "      <td>{'0': {'precision': 0.8767123287671232, 'recall': 0.8533333333333334, 'f1-score': 0.8648648648648648, 'support': 75}, '1': {'precision': 0.5925925925925926, 'recall': 0.64, 'f1-score': 0.6153846153846153, 'support': 25}, 'accuracy': 0.8, 'macro avg': {'precision': 0.7346524606798579, 'recall': 0.7466666666666667, 'f1-score': 0.74012474012474, 'support': 100}, 'weighted avg': {'precision': 0.8056823947234905, 'recall': 0.8, 'f1-score': 0.8024948024948024, 'support': 100}}</td>\n",
       "      <td>0.80</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>Hyperparams-oversampling-DecisionTreeClassifier</td>\n",
       "      <td>0.446059</td>\n",
       "      <td>{'0': {'precision': 0.9365079365079365, 'recall': 0.7866666666666666, 'f1-score': 0.8550724637681159, 'support': 75}, '1': {'precision': 0.5675675675675675, 'recall': 0.84, 'f1-score': 0.6774193548387097, 'support': 25}, 'accuracy': 0.8, 'macro avg': {'precision': 0.752037752037752, 'recall': 0.8133333333333332, 'f1-score': 0.7662459093034129, 'support': 100}, 'weighted avg': {'precision': 0.8442728442728443, 'recall': 0.8, 'f1-score': 0.8106591865357643, 'support': 100}}</td>\n",
       "      <td>0.80</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Hyperparams-smote-DecisionTreeClassifier</td>\n",
       "      <td>0.306592</td>\n",
       "      <td>{'0': {'precision': 0.95, 'recall': 0.76, 'f1-score': 0.8444444444444444, 'support': 75}, '1': {'precision': 0.55, 'recall': 0.88, 'f1-score': 0.676923076923077, 'support': 25}, 'accuracy': 0.79, 'macro avg': {'precision': 0.75, 'recall': 0.8200000000000001, 'f1-score': 0.7606837606837606, 'support': 100}, 'weighted avg': {'precision': 0.85, 'recall': 0.79, 'f1-score': 0.8025641025641026, 'support': 100}}</td>\n",
       "      <td>0.79</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Hyperparams-oversampling-XGBClassifier</td>\n",
       "      <td>0.822435</td>\n",
       "      <td>{'0': {'precision': 0.8533333333333334, 'recall': 0.8533333333333334, 'f1-score': 0.8533333333333335, 'support': 75}, '1': {'precision': 0.56, 'recall': 0.56, 'f1-score': 0.56, 'support': 25}, 'accuracy': 0.78, 'macro avg': {'precision': 0.7066666666666668, 'recall': 0.7066666666666668, 'f1-score': 0.7066666666666668, 'support': 100}, 'weighted avg': {'precision': 0.78, 'recall': 0.78, 'f1-score': 0.7800000000000001, 'support': 100}}</td>\n",
       "      <td>0.78</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Hyperparams-nonbalance-GradientBoostingClassifier</td>\n",
       "      <td>0.242505</td>\n",
       "      <td>{'0': {'precision': 0.7865168539325843, 'recall': 0.9333333333333333, 'f1-score': 0.853658536585366, 'support': 75}, '1': {'precision': 0.5454545454545454, 'recall': 0.24, 'f1-score': 0.3333333333333333, 'support': 25}, 'accuracy': 0.76, 'macro avg': {'precision': 0.6659856996935649, 'recall': 0.5866666666666667, 'f1-score': 0.5934959349593496, 'support': 100}, 'weighted avg': {'precision': 0.7262512768130747, 'recall': 0.76, 'f1-score': 0.7235772357723578, 'support': 100}}</td>\n",
       "      <td>0.76</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Hyperparams-nonbalance-XGBClassifier</td>\n",
       "      <td>0.842204</td>\n",
       "      <td>{'0': {'precision': 0.7659574468085106, 'recall': 0.96, 'f1-score': 0.8520710059171598, 'support': 75}, '1': {'precision': 0.5, 'recall': 0.12, 'f1-score': 0.1935483870967742, 'support': 25}, 'accuracy': 0.75, 'macro avg': {'precision': 0.6329787234042553, 'recall': 0.54, 'f1-score': 0.522809696506967, 'support': 100}, 'weighted avg': {'precision': 0.699468085106383, 'recall': 0.75, 'f1-score': 0.6874403512120634, 'support': 100}}</td>\n",
       "      <td>0.75</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                             model_name  training_time  \\\n",
       "4                       Hyperparams-smote-XGBClassifier       0.952034   \n",
       "9   Hyperparams-oversampling-GradientBoostingClassifier       0.445997   \n",
       "10          Hyperparams-oversampling-AdaBoostClassifier       0.507705   \n",
       "6                  Hyperparams-smote-AdaBoostClassifier       0.569004   \n",
       "2             Hyperparams-nonbalance-AdaBoostClassifier       0.252431   \n",
       "5          Hyperparams-smote-GradientBoostingClassifier       0.633001   \n",
       "3         Hyperparams-nonbalance-DecisionTreeClassifier       0.170997   \n",
       "11      Hyperparams-oversampling-DecisionTreeClassifier       0.446059   \n",
       "7              Hyperparams-smote-DecisionTreeClassifier       0.306592   \n",
       "8                Hyperparams-oversampling-XGBClassifier       0.822435   \n",
       "1     Hyperparams-nonbalance-GradientBoostingClassifier       0.242505   \n",
       "0                  Hyperparams-nonbalance-XGBClassifier       0.842204   \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       performance  \\\n",
       "4    {'0': {'precision': 0.8717948717948718, 'recall': 0.9066666666666666, 'f1-score': 0.8888888888888887, 'support': 75}, '1': {'precision': 0.6818181818181818, 'recall': 0.6, 'f1-score': 0.6382978723404256, 'support': 25}, 'accuracy': 0.83, 'macro avg': {'precision': 0.7768065268065267, 'recall': 0.7533333333333333, 'f1-score': 0.7635933806146571, 'support': 100}, 'weighted avg': {'precision': 0.8243006993006994, 'recall': 0.83, 'f1-score': 0.826241134751773, 'support': 100}}   \n",
       "9              {'0': {'precision': 0.9672131147540983, 'recall': 0.7866666666666666, 'f1-score': 0.8676470588235293, 'support': 75}, '1': {'precision': 0.5897435897435898, 'recall': 0.92, 'f1-score': 0.71875, 'support': 25}, 'accuracy': 0.82, 'macro avg': {'precision': 0.7784783522488441, 'recall': 0.8533333333333333, 'f1-score': 0.7931985294117647, 'support': 100}, 'weighted avg': {'precision': 0.8728457335014713, 'recall': 0.82, 'f1-score': 0.830422794117647, 'support': 100}}   \n",
       "10             {'0': {'precision': 0.9672131147540983, 'recall': 0.7866666666666666, 'f1-score': 0.8676470588235293, 'support': 75}, '1': {'precision': 0.5897435897435898, 'recall': 0.92, 'f1-score': 0.71875, 'support': 25}, 'accuracy': 0.82, 'macro avg': {'precision': 0.7784783522488441, 'recall': 0.8533333333333333, 'f1-score': 0.7931985294117647, 'support': 100}, 'weighted avg': {'precision': 0.8728457335014713, 'recall': 0.82, 'f1-score': 0.830422794117647, 'support': 100}}   \n",
       "6                 {'0': {'precision': 0.9384615384615385, 'recall': 0.8133333333333334, 'f1-score': 0.8714285714285714, 'support': 75}, '1': {'precision': 0.6, 'recall': 0.84, 'f1-score': 0.7000000000000001, 'support': 25}, 'accuracy': 0.82, 'macro avg': {'precision': 0.7692307692307692, 'recall': 0.8266666666666667, 'f1-score': 0.7857142857142858, 'support': 100}, 'weighted avg': {'precision': 0.8538461538461539, 'recall': 0.82, 'f1-score': 0.8285714285714286, 'support': 100}}   \n",
       "2    {'0': {'precision': 0.9242424242424242, 'recall': 0.8133333333333334, 'f1-score': 0.8652482269503545, 'support': 75}, '1': {'precision': 0.5882352941176471, 'recall': 0.8, 'f1-score': 0.6779661016949153, 'support': 25}, 'accuracy': 0.81, 'macro avg': {'precision': 0.7562388591800356, 'recall': 0.8066666666666666, 'f1-score': 0.771607164322635, 'support': 100}, 'weighted avg': {'precision': 0.8402406417112299, 'recall': 0.81, 'f1-score': 0.8184276956364946, 'support': 100}}   \n",
       "5                             {'0': {'precision': 0.9375, 'recall': 0.8, 'f1-score': 0.8633093525179856, 'support': 75}, '1': {'precision': 0.5833333333333334, 'recall': 0.84, 'f1-score': 0.6885245901639344, 'support': 25}, 'accuracy': 0.81, 'macro avg': {'precision': 0.7604166666666667, 'recall': 0.8200000000000001, 'f1-score': 0.7759169713409599, 'support': 100}, 'weighted avg': {'precision': 0.8489583333333333, 'recall': 0.81, 'f1-score': 0.8196131619294728, 'support': 100}}   \n",
       "3      {'0': {'precision': 0.8767123287671232, 'recall': 0.8533333333333334, 'f1-score': 0.8648648648648648, 'support': 75}, '1': {'precision': 0.5925925925925926, 'recall': 0.64, 'f1-score': 0.6153846153846153, 'support': 25}, 'accuracy': 0.8, 'macro avg': {'precision': 0.7346524606798579, 'recall': 0.7466666666666667, 'f1-score': 0.74012474012474, 'support': 100}, 'weighted avg': {'precision': 0.8056823947234905, 'recall': 0.8, 'f1-score': 0.8024948024948024, 'support': 100}}   \n",
       "11    {'0': {'precision': 0.9365079365079365, 'recall': 0.7866666666666666, 'f1-score': 0.8550724637681159, 'support': 75}, '1': {'precision': 0.5675675675675675, 'recall': 0.84, 'f1-score': 0.6774193548387097, 'support': 25}, 'accuracy': 0.8, 'macro avg': {'precision': 0.752037752037752, 'recall': 0.8133333333333332, 'f1-score': 0.7662459093034129, 'support': 100}, 'weighted avg': {'precision': 0.8442728442728443, 'recall': 0.8, 'f1-score': 0.8106591865357643, 'support': 100}}   \n",
       "7                                                                         {'0': {'precision': 0.95, 'recall': 0.76, 'f1-score': 0.8444444444444444, 'support': 75}, '1': {'precision': 0.55, 'recall': 0.88, 'f1-score': 0.676923076923077, 'support': 25}, 'accuracy': 0.79, 'macro avg': {'precision': 0.75, 'recall': 0.8200000000000001, 'f1-score': 0.7606837606837606, 'support': 100}, 'weighted avg': {'precision': 0.85, 'recall': 0.79, 'f1-score': 0.8025641025641026, 'support': 100}}   \n",
       "8                                            {'0': {'precision': 0.8533333333333334, 'recall': 0.8533333333333334, 'f1-score': 0.8533333333333335, 'support': 75}, '1': {'precision': 0.56, 'recall': 0.56, 'f1-score': 0.56, 'support': 25}, 'accuracy': 0.78, 'macro avg': {'precision': 0.7066666666666668, 'recall': 0.7066666666666668, 'f1-score': 0.7066666666666668, 'support': 100}, 'weighted avg': {'precision': 0.78, 'recall': 0.78, 'f1-score': 0.7800000000000001, 'support': 100}}   \n",
       "1   {'0': {'precision': 0.7865168539325843, 'recall': 0.9333333333333333, 'f1-score': 0.853658536585366, 'support': 75}, '1': {'precision': 0.5454545454545454, 'recall': 0.24, 'f1-score': 0.3333333333333333, 'support': 25}, 'accuracy': 0.76, 'macro avg': {'precision': 0.6659856996935649, 'recall': 0.5866666666666667, 'f1-score': 0.5934959349593496, 'support': 100}, 'weighted avg': {'precision': 0.7262512768130747, 'recall': 0.76, 'f1-score': 0.7235772357723578, 'support': 100}}   \n",
       "0                                               {'0': {'precision': 0.7659574468085106, 'recall': 0.96, 'f1-score': 0.8520710059171598, 'support': 75}, '1': {'precision': 0.5, 'recall': 0.12, 'f1-score': 0.1935483870967742, 'support': 25}, 'accuracy': 0.75, 'macro avg': {'precision': 0.6329787234042553, 'recall': 0.54, 'f1-score': 0.522809696506967, 'support': 100}, 'weighted avg': {'precision': 0.699468085106383, 'recall': 0.75, 'f1-score': 0.6874403512120634, 'support': 100}}   \n",
       "\n",
       "    f1_score_avg  \n",
       "4           0.83  \n",
       "9           0.82  \n",
       "10          0.82  \n",
       "6           0.82  \n",
       "2           0.81  \n",
       "5           0.81  \n",
       "3           0.80  \n",
       "11          0.80  \n",
       "7           0.79  \n",
       "8           0.78  \n",
       "1           0.76  \n",
       "0           0.75  "
      ]
     },
     "execution_count": 91,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.set_option('display.max_colwidth', None)\n",
    "# pd.reset_option('display.max_colwidth')\n",
    "pd.DataFrame(training_log_valid)[['model_name',\n",
    "                                 'training_time',\n",
    "                                 'performance',\n",
    "                                 'f1_score_avg']].sort_values([\"f1_score_avg\", \"training_time\"], ascending = [False, True])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 7. Test Data"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 7.1 Create ROC Curve based on Test Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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",
      "text/plain": [
       "<Figure size 720x576 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "## Create Proba for all function\n",
    "y_valid_xgb_over_proba = xgb_over.predict_proba(x_test)\n",
    "y_valid_xgb_smote_proba = xgb_smote.predict_proba(x_test)\n",
    "y_valid_dt_proba = dt_tree.predict_proba(x_test)\n",
    "y_valid_ada_over_proba = ada_over.predict_proba(x_test)\n",
    "y_valid_gb_pred_over_proba = grad_tree.predict_proba(x_test)\n",
    "y_valid_gb_pred_smote_proba = grad_tree_smote.predict_proba(x_test)\n",
    "\n",
    "## ROC CURVE Function for create FPR & TPR\n",
    "fpr_xgb_over, tpr_xgb_over, threshold_xgb_over = roc_curve(y_test,\n",
    "                                         y_valid_xgb_over_proba[:, 1])\n",
    "fpr_xgb_smote, tpr_xgb_smote, threshold_xgb_smote = roc_curve(y_test,\n",
    "                                         y_valid_xgb_smote_proba[:, 1])\n",
    "fpr_dt, tpr_dt, threshold_dt = roc_curve(y_test,\n",
    "                                        y_valid_dt_proba[:, 1])\n",
    "fpr_ada_over, tpr_ada_over, threshold_ada_over = roc_curve(y_test,\n",
    "                                                           y_valid_ada_over_proba[:, 1])\n",
    "fpr_gb_over, tpr_gb_over, threshold_gb_over = roc_curve(y_test,\n",
    "                                         y_valid_gb_pred_over_proba[:, 1])\n",
    "\n",
    "fpr_gb_smote, tpr_gb_smote, threshold_gb_smote = roc_curve(y_test,\n",
    "                                         y_valid_gb_pred_smote_proba[:, 1])\n",
    "\n",
    "roc_auc_xgb_over = auc(fpr_xgb_over, tpr_xgb_over)\n",
    "roc_auc_xgb_smote = auc(fpr_xgb_smote, tpr_xgb_smote)\n",
    "roc_auc_dt = auc(fpr_dt, tpr_dt)\n",
    "roc_auc_ada_over = auc(fpr_ada_over, tpr_ada_over)\n",
    "roc_auc_gb_over = auc(fpr_gb_over, tpr_gb_over)\n",
    "roc_auc_gb_smote = auc(fpr_gb_smote, tpr_gb_smote)\n",
    "\n",
    "\n",
    "plt.subplots(figsize=(10,8))\n",
    "plt.title('Receiver Operating Characteristic')\n",
    "\n",
    "plt.plot(fpr_gb_over, tpr_gb_over, 'b', label = 'AUC GradientBoost Oversampling = %0.2f' % roc_auc_gb_over)\n",
    "plt.plot(fpr_gb_smote, tpr_gb_smote, 'g', label = 'AUC GradientBoost Smote = %0.2f' % roc_auc_gb_smote)\n",
    "plt.plot(fpr_xgb_over, tpr_xgb_over, 'k', label = 'AUC XGBoost Oversampling = %0.2f' % roc_auc_xgb_over)\n",
    "plt.plot(fpr_xgb_smote, tpr_xgb_smote, 'r', label = 'AUC XGBoost Smote = %0.2f' % roc_auc_xgb_smote)\n",
    "plt.plot(fpr_dt, tpr_dt, 'y', label = 'AUC DecisionTree = %0.2f' % roc_auc_dt)\n",
    "plt.plot(fpr_ada_over, tpr_ada_over, 'c', label = 'AUC AdaBoost smote = %0.2f' % roc_auc_ada_over)\n",
    "\n",
    "plt.legend(loc = 'lower right')\n",
    "plt.plot([0, 1], [0, 1],'r--')\n",
    "plt.xlim([0, 1])\n",
    "plt.ylim([0, 1])\n",
    "plt.ylabel('True Positive Rate')\n",
    "plt.xlabel('False Positive Rate')\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 7.2 Create Function to Eval Test Data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 63,
   "metadata": {},
   "outputs": [],
   "source": [
    "def train_eval_test(configuration_model: str, params: dict, hyperparams_model: list = None):\n",
    "\n",
    "    # Variabel to store trained models\n",
    "    list_of_trained_model = dict()\n",
    "\n",
    "    # Create log template\n",
    "    training_log = training_log_template()\n",
    "\n",
    "    # Training for every data configuration\n",
    "    for config_data in x_train:\n",
    "        # Debug message\n",
    "        util.print_debug(\"Training model based on configuration data: {}\".format(config_data))\n",
    "\n",
    "        # Create model objects\n",
    "        if hyperparams_model == None:\n",
    "            list_of_model = create_model_object(params)\n",
    "        else:\n",
    "            list_of_model = copy.deepcopy(hyperparams_model)\n",
    "\n",
    "        # Variabel to store tained model\n",
    "        trained_model = list()\n",
    "\n",
    "        # Load train data based on its configuration\n",
    "        x_train_data = x_train[config_data]\n",
    "        y_train_data = y_train[config_data]\n",
    "\n",
    "        # Train each model by current dataset configuration\n",
    "        for model in list_of_model:\n",
    "            # Debug message\n",
    "            util.print_debug(\"Training model: {}\".format(model[\"model_name\"]))\n",
    "\n",
    "            # Training\n",
    "            training_time = util.time_stamp()\n",
    "            model[\"model_object\"].fit(x_train_data, y_train_data)\n",
    "            training_time = (util.time_stamp() - training_time).total_seconds()\n",
    "\n",
    "            # Debug message\n",
    "            util.print_debug(\"Evalutaing model: {}\".format(model[\"model_name\"]))\n",
    "\n",
    "            # Evaluation\n",
    "            y_predict = model[\"model_object\"].predict(x_test)\n",
    "            performance = classification_report(y_test, y_predict, output_dict = True)\n",
    "\n",
    "            # Debug message\n",
    "            util.print_debug(\"Logging: {}\".format(model[\"model_name\"]))\n",
    "\n",
    "            # Create UID\n",
    "            uid = hashlib.md5(str(training_time).encode()).hexdigest()\n",
    "\n",
    "            # Assign model's UID\n",
    "            model[\"model_uid\"] = uid\n",
    "\n",
    "            # Create training log data\n",
    "            training_log[\"model_name\"].append(\"{}-{}-{}\".format(configuration_model, config_data, model[\"model_name\"]))\n",
    "            training_log[\"model_uid\"].append(uid)\n",
    "            training_log[\"training_time\"].append(training_time)\n",
    "            training_log[\"training_date\"].append(util.time_stamp())\n",
    "            training_log[\"performance\"].append(performance)\n",
    "            training_log[\"f1_score_avg\"].append(performance[\"weighted avg\"][\"recall\"])\n",
    "            training_log[\"data_configurations\"].append(config_data)\n",
    "\n",
    "            # Collect current trained model\n",
    "            trained_model.append(copy.deepcopy(model))\n",
    "\n",
    "            # Debug message\n",
    "            util.print_debug(\"Model {} has been trained for configuration data {}.\".format(model[\"model_name\"], config_data))\n",
    "        \n",
    "        # Collect current trained list of model\n",
    "        list_of_trained_model[config_data] = copy.deepcopy(trained_model)\n",
    "    \n",
    "    \n",
    "    # Debug message\n",
    "    util.print_debug(\"All combination models and configuration data has been trained.\")\n",
    "    \n",
    "    # Return list trained model\n",
    "    return list_of_trained_model, training_log"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 7.3 Training Dataset and Update Training Logs"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 75,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2023-01-07 22:16:01.574957 creating training log template.\n",
      "2023-01-07 22:16:01.575956 Training log template created.\n",
      "2023-01-07 22:16:01.575956 Training model based on configuration data: nonbalance\n",
      "2023-01-07 22:16:01.577959 Training model: XGBClassifier\n",
      "2023-01-07 22:16:02.180261 Evalutaing model: XGBClassifier\n",
      "2023-01-07 22:16:02.232262 Logging: XGBClassifier\n",
      "2023-01-07 22:16:02.244276 Model XGBClassifier has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:16:02.244276 Training model: GradientBoostingClassifier\n",
      "2023-01-07 22:16:02.539229 Evalutaing model: GradientBoostingClassifier\n",
      "2023-01-07 22:16:02.562647 Logging: GradientBoostingClassifier\n",
      "2023-01-07 22:16:02.569643 Model GradientBoostingClassifier has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:16:02.569643 Training model: AdaBoostClassifier\n",
      "2023-01-07 22:16:02.815179 Evalutaing model: AdaBoostClassifier\n",
      "2023-01-07 22:16:02.840880 Logging: AdaBoostClassifier\n",
      "2023-01-07 22:16:02.844880 Model AdaBoostClassifier has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:16:02.845875 Training model: DecisionTreeClassifier\n",
      "2023-01-07 22:16:03.026638 Evalutaing model: DecisionTreeClassifier\n",
      "2023-01-07 22:16:03.051639 Logging: DecisionTreeClassifier\n",
      "2023-01-07 22:16:03.054641 Model DecisionTreeClassifier has been trained for configuration data nonbalance.\n",
      "2023-01-07 22:16:03.082638 Training model based on configuration data: smote\n",
      "2023-01-07 22:16:03.085638 Training model: XGBClassifier\n",
      "2023-01-07 22:16:03.985798 Evalutaing model: XGBClassifier\n",
      "2023-01-07 22:16:04.026795 Logging: XGBClassifier\n",
      "2023-01-07 22:16:04.033796 Model XGBClassifier has been trained for configuration data smote.\n",
      "2023-01-07 22:16:04.033796 Training model: GradientBoostingClassifier\n",
      "2023-01-07 22:16:04.543477 Evalutaing model: GradientBoostingClassifier\n",
      "2023-01-07 22:16:04.571479 Logging: GradientBoostingClassifier\n",
      "2023-01-07 22:16:04.579473 Model GradientBoostingClassifier has been trained for configuration data smote.\n",
      "2023-01-07 22:16:04.580475 Training model: AdaBoostClassifier\n",
      "2023-01-07 22:16:05.082311 Evalutaing model: AdaBoostClassifier\n",
      "2023-01-07 22:16:05.114311 Logging: AdaBoostClassifier\n",
      "2023-01-07 22:16:05.120311 Model AdaBoostClassifier has been trained for configuration data smote.\n",
      "2023-01-07 22:16:05.121312 Training model: DecisionTreeClassifier\n",
      "2023-01-07 22:16:05.455212 Evalutaing model: DecisionTreeClassifier\n",
      "2023-01-07 22:16:05.476772 Logging: DecisionTreeClassifier\n",
      "2023-01-07 22:16:05.479775 Model DecisionTreeClassifier has been trained for configuration data smote.\n",
      "2023-01-07 22:16:05.496357 Training model based on configuration data: oversampling\n",
      "2023-01-07 22:16:05.499358 Training model: XGBClassifier\n",
      "2023-01-07 22:16:06.418186 Evalutaing model: XGBClassifier\n",
      "2023-01-07 22:16:06.462186 Logging: XGBClassifier\n",
      "2023-01-07 22:16:06.470186 Model XGBClassifier has been trained for configuration data oversampling.\n",
      "2023-01-07 22:16:06.470186 Training model: GradientBoostingClassifier\n",
      "2023-01-07 22:16:06.941990 Evalutaing model: GradientBoostingClassifier\n",
      "2023-01-07 22:16:06.993974 Logging: GradientBoostingClassifier\n",
      "2023-01-07 22:16:06.999974 Model GradientBoostingClassifier has been trained for configuration data oversampling.\n",
      "2023-01-07 22:16:07.002974 Training model: AdaBoostClassifier\n",
      "2023-01-07 22:16:07.442014 Evalutaing model: AdaBoostClassifier\n",
      "2023-01-07 22:16:07.469017 Logging: AdaBoostClassifier\n",
      "2023-01-07 22:16:07.475012 Model AdaBoostClassifier has been trained for configuration data oversampling.\n",
      "2023-01-07 22:16:07.475012 Training model: DecisionTreeClassifier\n",
      "2023-01-07 22:16:07.799011 Evalutaing model: DecisionTreeClassifier\n",
      "2023-01-07 22:16:07.821442 Logging: DecisionTreeClassifier\n",
      "2023-01-07 22:16:07.823442 Model DecisionTreeClassifier has been trained for configuration data oversampling.\n",
      "2023-01-07 22:16:07.841441 All combination models and configuration data has been trained.\n"
     ]
    }
   ],
   "source": [
    "list_of_trained_model_test, \\\n",
    "training_log_test = train_eval_test(\"Hyperparams\", \n",
    "                                    config_data,\n",
    "                                    hyper_params_tuning(['XGBClassifier',\n",
    "                                                          'GradientBoostingClassifier', \n",
    "                                                          'AdaBoostClassifier',\n",
    "                                                          'DecisionTreeClassifier']))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 7.4 Check Best Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model_name</th>\n",
       "      <th>training_time</th>\n",
       "      <th>performance</th>\n",
       "      <th>f1_score_avg</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Hyperparams-smote-GradientBoostingClassifier</td>\n",
       "      <td>0.509681</td>\n",
       "      <td>{'0': {'precision': 0.9333333333333333, 'recall': 0.9210526315789473, 'f1-score': 0.9271523178807947, 'support': 76}, '1': {'precision': 0.76, 'recall': 0.7916666666666666, 'f1-score': 0.7755102040816326, 'support': 24}, 'accuracy': 0.89, 'macro avg': {'precision': 0.8466666666666667, 'recall': 0.8563596491228069, 'f1-score': 0.8513312609812136, 'support': 100}, 'weighted avg': {'precision': 0.8917333333333335, 'recall': 0.89, 'f1-score': 0.8907582105689958, 'support': 100}}</td>\n",
       "      <td>0.89</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Hyperparams-oversampling-AdaBoostClassifier</td>\n",
       "      <td>0.439040</td>\n",
       "      <td>{'0': {'precision': 0.9315068493150684, 'recall': 0.8947368421052632, 'f1-score': 0.912751677852349, 'support': 76}, '1': {'precision': 0.7037037037037037, 'recall': 0.7916666666666666, 'f1-score': 0.7450980392156864, 'support': 24}, 'accuracy': 0.87, 'macro avg': {'precision': 0.8176052765093861, 'recall': 0.8432017543859649, 'f1-score': 0.8289248585340177, 'support': 100}, 'weighted avg': {'precision': 0.8768340943683408, 'recall': 0.87, 'f1-score': 0.87251480457955, 'support': 100}}</td>\n",
       "      <td>0.87</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>Hyperparams-oversampling-GradientBoostingClassifier</td>\n",
       "      <td>0.470805</td>\n",
       "      <td>{'0': {'precision': 0.9315068493150684, 'recall': 0.8947368421052632, 'f1-score': 0.912751677852349, 'support': 76}, '1': {'precision': 0.7037037037037037, 'recall': 0.7916666666666666, 'f1-score': 0.7450980392156864, 'support': 24}, 'accuracy': 0.87, 'macro avg': {'precision': 0.8176052765093861, 'recall': 0.8432017543859649, 'f1-score': 0.8289248585340177, 'support': 100}, 'weighted avg': {'precision': 0.8768340943683408, 'recall': 0.87, 'f1-score': 0.87251480457955, 'support': 100}}</td>\n",
       "      <td>0.87</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Hyperparams-smote-DecisionTreeClassifier</td>\n",
       "      <td>0.333900</td>\n",
       "      <td>{'0': {'precision': 0.918918918918919, 'recall': 0.8947368421052632, 'f1-score': 0.9066666666666667, 'support': 76}, '1': {'precision': 0.6923076923076923, 'recall': 0.75, 'f1-score': 0.7199999999999999, 'support': 24}, 'accuracy': 0.86, 'macro avg': {'precision': 0.8056133056133057, 'recall': 0.8223684210526316, 'f1-score': 0.8133333333333332, 'support': 100}, 'weighted avg': {'precision': 0.8645322245322246, 'recall': 0.86, 'f1-score': 0.8618666666666667, 'support': 100}}</td>\n",
       "      <td>0.86</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Hyperparams-smote-AdaBoostClassifier</td>\n",
       "      <td>0.501836</td>\n",
       "      <td>{'0': {'precision': 0.875, 'recall': 0.9210526315789473, 'f1-score': 0.8974358974358975, 'support': 76}, '1': {'precision': 0.7, 'recall': 0.5833333333333334, 'f1-score': 0.6363636363636365, 'support': 24}, 'accuracy': 0.84, 'macro avg': {'precision': 0.7875, 'recall': 0.7521929824561404, 'f1-score': 0.766899766899767, 'support': 100}, 'weighted avg': {'precision': 0.833, 'recall': 0.84, 'f1-score': 0.8347785547785549, 'support': 100}}</td>\n",
       "      <td>0.84</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Hyperparams-nonbalance-AdaBoostClassifier</td>\n",
       "      <td>0.243534</td>\n",
       "      <td>{'0': {'precision': 0.8641975308641975, 'recall': 0.9210526315789473, 'f1-score': 0.89171974522293, 'support': 76}, '1': {'precision': 0.6842105263157895, 'recall': 0.5416666666666666, 'f1-score': 0.6046511627906976, 'support': 24}, 'accuracy': 0.83, 'macro avg': {'precision': 0.7742040285899935, 'recall': 0.7313596491228069, 'f1-score': 0.7481854540068138, 'support': 100}, 'weighted avg': {'precision': 0.8210006497725796, 'recall': 0.83, 'f1-score': 0.8228232854391941, 'support': 100}}</td>\n",
       "      <td>0.83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Hyperparams-smote-XGBClassifier</td>\n",
       "      <td>0.899158</td>\n",
       "      <td>{'0': {'precision': 0.8641975308641975, 'recall': 0.9210526315789473, 'f1-score': 0.89171974522293, 'support': 76}, '1': {'precision': 0.6842105263157895, 'recall': 0.5416666666666666, 'f1-score': 0.6046511627906976, 'support': 24}, 'accuracy': 0.83, 'macro avg': {'precision': 0.7742040285899935, 'recall': 0.7313596491228069, 'f1-score': 0.7481854540068138, 'support': 100}, 'weighted avg': {'precision': 0.8210006497725796, 'recall': 0.83, 'f1-score': 0.8228232854391941, 'support': 100}}</td>\n",
       "      <td>0.83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Hyperparams-nonbalance-DecisionTreeClassifier</td>\n",
       "      <td>0.180763</td>\n",
       "      <td>{'0': {'precision': 0.8625, 'recall': 0.9078947368421053, 'f1-score': 0.8846153846153847, 'support': 76}, '1': {'precision': 0.65, 'recall': 0.5416666666666666, 'f1-score': 0.5909090909090908, 'support': 24}, 'accuracy': 0.82, 'macro avg': {'precision': 0.7562500000000001, 'recall': 0.7247807017543859, 'f1-score': 0.7377622377622377, 'support': 100}, 'weighted avg': {'precision': 0.8115000000000001, 'recall': 0.82, 'f1-score': 0.8141258741258741, 'support': 100}}</td>\n",
       "      <td>0.82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Hyperparams-oversampling-XGBClassifier</td>\n",
       "      <td>0.917828</td>\n",
       "      <td>{'0': {'precision': 0.8717948717948718, 'recall': 0.8947368421052632, 'f1-score': 0.8831168831168831, 'support': 76}, '1': {'precision': 0.6363636363636364, 'recall': 0.5833333333333334, 'f1-score': 0.6086956521739131, 'support': 24}, 'accuracy': 0.82, 'macro avg': {'precision': 0.754079254079254, 'recall': 0.7390350877192983, 'f1-score': 0.745906267645398, 'support': 100}, 'weighted avg': {'precision': 0.8152913752913753, 'recall': 0.82, 'f1-score': 0.8172557876905703, 'support': 100}}</td>\n",
       "      <td>0.82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>Hyperparams-oversampling-DecisionTreeClassifier</td>\n",
       "      <td>0.323999</td>\n",
       "      <td>{'0': {'precision': 0.9, 'recall': 0.8289473684210527, 'f1-score': 0.8630136986301371, 'support': 76}, '1': {'precision': 0.5666666666666667, 'recall': 0.7083333333333334, 'f1-score': 0.6296296296296297, 'support': 24}, 'accuracy': 0.8, 'macro avg': {'precision': 0.7333333333333334, 'recall': 0.7686403508771931, 'f1-score': 0.7463216641298833, 'support': 100}, 'weighted avg': {'precision': 0.82, 'recall': 0.8, 'f1-score': 0.8070015220700154, 'support': 100}}</td>\n",
       "      <td>0.80</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Hyperparams-nonbalance-GradientBoostingClassifier</td>\n",
       "      <td>0.292970</td>\n",
       "      <td>{'0': {'precision': 0.7659574468085106, 'recall': 0.9473684210526315, 'f1-score': 0.8470588235294116, 'support': 76}, '1': {'precision': 0.3333333333333333, 'recall': 0.08333333333333333, 'f1-score': 0.13333333333333333, 'support': 24}, 'accuracy': 0.74, 'macro avg': {'precision': 0.549645390070922, 'recall': 0.5153508771929824, 'f1-score': 0.4901960784313725, 'support': 100}, 'weighted avg': {'precision': 0.6621276595744681, 'recall': 0.74, 'f1-score': 0.6757647058823528, 'support': 100}}</td>\n",
       "      <td>0.74</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Hyperparams-nonbalance-XGBClassifier</td>\n",
       "      <td>0.601304</td>\n",
       "      <td>{'0': {'precision': 0.7551020408163265, 'recall': 0.9736842105263158, 'f1-score': 0.8505747126436782, 'support': 76}, '1': {'precision': 0.0, 'recall': 0.0, 'f1-score': 0.0, 'support': 24}, 'accuracy': 0.74, 'macro avg': {'precision': 0.37755102040816324, 'recall': 0.4868421052631579, 'f1-score': 0.4252873563218391, 'support': 100}, 'weighted avg': {'precision': 0.5738775510204082, 'recall': 0.74, 'f1-score': 0.6464367816091955, 'support': 100}}</td>\n",
       "      <td>0.74</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                             model_name  training_time  \\\n",
       "5          Hyperparams-smote-GradientBoostingClassifier       0.509681   \n",
       "10          Hyperparams-oversampling-AdaBoostClassifier       0.439040   \n",
       "9   Hyperparams-oversampling-GradientBoostingClassifier       0.470805   \n",
       "7              Hyperparams-smote-DecisionTreeClassifier       0.333900   \n",
       "6                  Hyperparams-smote-AdaBoostClassifier       0.501836   \n",
       "2             Hyperparams-nonbalance-AdaBoostClassifier       0.243534   \n",
       "4                       Hyperparams-smote-XGBClassifier       0.899158   \n",
       "3         Hyperparams-nonbalance-DecisionTreeClassifier       0.180763   \n",
       "8                Hyperparams-oversampling-XGBClassifier       0.917828   \n",
       "11      Hyperparams-oversampling-DecisionTreeClassifier       0.323999   \n",
       "1     Hyperparams-nonbalance-GradientBoostingClassifier       0.292970   \n",
       "0                  Hyperparams-nonbalance-XGBClassifier       0.601304   \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       performance  \\\n",
       "5                  {'0': {'precision': 0.9333333333333333, 'recall': 0.9210526315789473, 'f1-score': 0.9271523178807947, 'support': 76}, '1': {'precision': 0.76, 'recall': 0.7916666666666666, 'f1-score': 0.7755102040816326, 'support': 24}, 'accuracy': 0.89, 'macro avg': {'precision': 0.8466666666666667, 'recall': 0.8563596491228069, 'f1-score': 0.8513312609812136, 'support': 100}, 'weighted avg': {'precision': 0.8917333333333335, 'recall': 0.89, 'f1-score': 0.8907582105689958, 'support': 100}}   \n",
       "10      {'0': {'precision': 0.9315068493150684, 'recall': 0.8947368421052632, 'f1-score': 0.912751677852349, 'support': 76}, '1': {'precision': 0.7037037037037037, 'recall': 0.7916666666666666, 'f1-score': 0.7450980392156864, 'support': 24}, 'accuracy': 0.87, 'macro avg': {'precision': 0.8176052765093861, 'recall': 0.8432017543859649, 'f1-score': 0.8289248585340177, 'support': 100}, 'weighted avg': {'precision': 0.8768340943683408, 'recall': 0.87, 'f1-score': 0.87251480457955, 'support': 100}}   \n",
       "9       {'0': {'precision': 0.9315068493150684, 'recall': 0.8947368421052632, 'f1-score': 0.912751677852349, 'support': 76}, '1': {'precision': 0.7037037037037037, 'recall': 0.7916666666666666, 'f1-score': 0.7450980392156864, 'support': 24}, 'accuracy': 0.87, 'macro avg': {'precision': 0.8176052765093861, 'recall': 0.8432017543859649, 'f1-score': 0.8289248585340177, 'support': 100}, 'weighted avg': {'precision': 0.8768340943683408, 'recall': 0.87, 'f1-score': 0.87251480457955, 'support': 100}}   \n",
       "7                   {'0': {'precision': 0.918918918918919, 'recall': 0.8947368421052632, 'f1-score': 0.9066666666666667, 'support': 76}, '1': {'precision': 0.6923076923076923, 'recall': 0.75, 'f1-score': 0.7199999999999999, 'support': 24}, 'accuracy': 0.86, 'macro avg': {'precision': 0.8056133056133057, 'recall': 0.8223684210526316, 'f1-score': 0.8133333333333332, 'support': 100}, 'weighted avg': {'precision': 0.8645322245322246, 'recall': 0.86, 'f1-score': 0.8618666666666667, 'support': 100}}   \n",
       "6                                                          {'0': {'precision': 0.875, 'recall': 0.9210526315789473, 'f1-score': 0.8974358974358975, 'support': 76}, '1': {'precision': 0.7, 'recall': 0.5833333333333334, 'f1-score': 0.6363636363636365, 'support': 24}, 'accuracy': 0.84, 'macro avg': {'precision': 0.7875, 'recall': 0.7521929824561404, 'f1-score': 0.766899766899767, 'support': 100}, 'weighted avg': {'precision': 0.833, 'recall': 0.84, 'f1-score': 0.8347785547785549, 'support': 100}}   \n",
       "2      {'0': {'precision': 0.8641975308641975, 'recall': 0.9210526315789473, 'f1-score': 0.89171974522293, 'support': 76}, '1': {'precision': 0.6842105263157895, 'recall': 0.5416666666666666, 'f1-score': 0.6046511627906976, 'support': 24}, 'accuracy': 0.83, 'macro avg': {'precision': 0.7742040285899935, 'recall': 0.7313596491228069, 'f1-score': 0.7481854540068138, 'support': 100}, 'weighted avg': {'precision': 0.8210006497725796, 'recall': 0.83, 'f1-score': 0.8228232854391941, 'support': 100}}   \n",
       "4      {'0': {'precision': 0.8641975308641975, 'recall': 0.9210526315789473, 'f1-score': 0.89171974522293, 'support': 76}, '1': {'precision': 0.6842105263157895, 'recall': 0.5416666666666666, 'f1-score': 0.6046511627906976, 'support': 24}, 'accuracy': 0.83, 'macro avg': {'precision': 0.7742040285899935, 'recall': 0.7313596491228069, 'f1-score': 0.7481854540068138, 'support': 100}, 'weighted avg': {'precision': 0.8210006497725796, 'recall': 0.83, 'f1-score': 0.8228232854391941, 'support': 100}}   \n",
       "3                              {'0': {'precision': 0.8625, 'recall': 0.9078947368421053, 'f1-score': 0.8846153846153847, 'support': 76}, '1': {'precision': 0.65, 'recall': 0.5416666666666666, 'f1-score': 0.5909090909090908, 'support': 24}, 'accuracy': 0.82, 'macro avg': {'precision': 0.7562500000000001, 'recall': 0.7247807017543859, 'f1-score': 0.7377622377622377, 'support': 100}, 'weighted avg': {'precision': 0.8115000000000001, 'recall': 0.82, 'f1-score': 0.8141258741258741, 'support': 100}}   \n",
       "8      {'0': {'precision': 0.8717948717948718, 'recall': 0.8947368421052632, 'f1-score': 0.8831168831168831, 'support': 76}, '1': {'precision': 0.6363636363636364, 'recall': 0.5833333333333334, 'f1-score': 0.6086956521739131, 'support': 24}, 'accuracy': 0.82, 'macro avg': {'precision': 0.754079254079254, 'recall': 0.7390350877192983, 'f1-score': 0.745906267645398, 'support': 100}, 'weighted avg': {'precision': 0.8152913752913753, 'recall': 0.82, 'f1-score': 0.8172557876905703, 'support': 100}}   \n",
       "11                                  {'0': {'precision': 0.9, 'recall': 0.8289473684210527, 'f1-score': 0.8630136986301371, 'support': 76}, '1': {'precision': 0.5666666666666667, 'recall': 0.7083333333333334, 'f1-score': 0.6296296296296297, 'support': 24}, 'accuracy': 0.8, 'macro avg': {'precision': 0.7333333333333334, 'recall': 0.7686403508771931, 'f1-score': 0.7463216641298833, 'support': 100}, 'weighted avg': {'precision': 0.82, 'recall': 0.8, 'f1-score': 0.8070015220700154, 'support': 100}}   \n",
       "1   {'0': {'precision': 0.7659574468085106, 'recall': 0.9473684210526315, 'f1-score': 0.8470588235294116, 'support': 76}, '1': {'precision': 0.3333333333333333, 'recall': 0.08333333333333333, 'f1-score': 0.13333333333333333, 'support': 24}, 'accuracy': 0.74, 'macro avg': {'precision': 0.549645390070922, 'recall': 0.5153508771929824, 'f1-score': 0.4901960784313725, 'support': 100}, 'weighted avg': {'precision': 0.6621276595744681, 'recall': 0.74, 'f1-score': 0.6757647058823528, 'support': 100}}   \n",
       "0                                                {'0': {'precision': 0.7551020408163265, 'recall': 0.9736842105263158, 'f1-score': 0.8505747126436782, 'support': 76}, '1': {'precision': 0.0, 'recall': 0.0, 'f1-score': 0.0, 'support': 24}, 'accuracy': 0.74, 'macro avg': {'precision': 0.37755102040816324, 'recall': 0.4868421052631579, 'f1-score': 0.4252873563218391, 'support': 100}, 'weighted avg': {'precision': 0.5738775510204082, 'recall': 0.74, 'f1-score': 0.6464367816091955, 'support': 100}}   \n",
       "\n",
       "    f1_score_avg  \n",
       "5           0.89  \n",
       "10          0.87  \n",
       "9           0.87  \n",
       "7           0.86  \n",
       "6           0.84  \n",
       "2           0.83  \n",
       "4           0.83  \n",
       "3           0.82  \n",
       "8           0.82  \n",
       "11          0.80  \n",
       "1           0.74  \n",
       "0           0.74  "
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.set_option('display.max_colwidth', None)\n",
    "# pd.reset_option('display.max_colwidth')\n",
    "pd.DataFrame(training_log_test)[['model_name',\n",
    "                                 'training_time',\n",
    "                                 'performance',\n",
    "                                 'f1_score_avg']].sort_values([\"f1_score_avg\", \"training_time\"], ascending = [False, True])"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 8. Save Production Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "2023-01-07 23:21:23.121474 Choosing model by metrics score.\n",
      "2023-01-07 23:21:23.121474 Converting training log type of data from dict to dataframe.\n",
      "2023-01-07 23:21:23.128473 Trying to load previous production model.\n",
      "2023-01-07 23:21:23.176474 Previous production model loaded.\n",
      "2023-01-07 23:21:23.176474 Loading validation data.\n",
      "2023-01-07 23:21:23.176474 Checking compatibilty previous production model's input with current train data's features.\n",
      "2023-01-07 23:21:23.177473 Features compatible.\n",
      "2023-01-07 23:21:23.177473 Reassesing previous model performance using current validation data.\n",
      "2023-01-07 23:21:23.206476 Assessing complete.\n",
      "2023-01-07 23:21:23.206476 Storing new metrics data to previous model structure.\n",
      "2023-01-07 23:21:23.206476 Adding previous model data to current training log and list of model\n",
      "2023-01-07 23:21:23.216478 Sorting training log by f1 macro avg and training time.\n",
      "2023-01-07 23:21:23.221474 Searching model data based on sorted training log.\n",
      "2023-01-07 23:21:23.258376 Model chosen.\n"
     ]
    }
   ],
   "source": [
    "model_test, \\\n",
    "production_model_log_test, \\\n",
    "training_logs_test = get_production_model(list_of_trained_model_test, \n",
    "                                          training_log_test, \n",
    "                                          config_data)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 8.1 Show Best Final Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {
    "scrolled": false
   },
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>model_name</th>\n",
       "      <th>training_time</th>\n",
       "      <th>performance</th>\n",
       "      <th>f1_score_avg</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>Hyperparams-smote-GradientBoostingClassifier</td>\n",
       "      <td>0.509681</td>\n",
       "      <td>{'0': {'precision': 0.9333333333333333, 'recall': 0.9210526315789473, 'f1-score': 0.9271523178807947, 'support': 76}, '1': {'precision': 0.76, 'recall': 0.7916666666666666, 'f1-score': 0.7755102040816326, 'support': 24}, 'accuracy': 0.89, 'macro avg': {'precision': 0.8466666666666667, 'recall': 0.8563596491228069, 'f1-score': 0.8513312609812136, 'support': 100}, 'weighted avg': {'precision': 0.8917333333333335, 'recall': 0.89, 'f1-score': 0.8907582105689958, 'support': 100}}</td>\n",
       "      <td>0.89</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>Hyperparams-oversampling-GradientBoostingClassifier</td>\n",
       "      <td>0.470805</td>\n",
       "      <td>{'0': {'precision': 0.9315068493150684, 'recall': 0.8947368421052632, 'f1-score': 0.912751677852349, 'support': 76}, '1': {'precision': 0.7037037037037037, 'recall': 0.7916666666666666, 'f1-score': 0.7450980392156864, 'support': 24}, 'accuracy': 0.87, 'macro avg': {'precision': 0.8176052765093861, 'recall': 0.8432017543859649, 'f1-score': 0.8289248585340177, 'support': 100}, 'weighted avg': {'precision': 0.8768340943683408, 'recall': 0.87, 'f1-score': 0.87251480457955, 'support': 100}}</td>\n",
       "      <td>0.87</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>Hyperparams-oversampling-AdaBoostClassifier</td>\n",
       "      <td>0.439040</td>\n",
       "      <td>{'0': {'precision': 0.9315068493150684, 'recall': 0.8947368421052632, 'f1-score': 0.912751677852349, 'support': 76}, '1': {'precision': 0.7037037037037037, 'recall': 0.7916666666666666, 'f1-score': 0.7450980392156864, 'support': 24}, 'accuracy': 0.87, 'macro avg': {'precision': 0.8176052765093861, 'recall': 0.8432017543859649, 'f1-score': 0.8289248585340177, 'support': 100}, 'weighted avg': {'precision': 0.8768340943683408, 'recall': 0.87, 'f1-score': 0.87251480457955, 'support': 100}}</td>\n",
       "      <td>0.87</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>Hyperparams-smote-DecisionTreeClassifier</td>\n",
       "      <td>0.333900</td>\n",
       "      <td>{'0': {'precision': 0.918918918918919, 'recall': 0.8947368421052632, 'f1-score': 0.9066666666666667, 'support': 76}, '1': {'precision': 0.6923076923076923, 'recall': 0.75, 'f1-score': 0.7199999999999999, 'support': 24}, 'accuracy': 0.86, 'macro avg': {'precision': 0.8056133056133057, 'recall': 0.8223684210526316, 'f1-score': 0.8133333333333332, 'support': 100}, 'weighted avg': {'precision': 0.8645322245322246, 'recall': 0.86, 'f1-score': 0.8618666666666667, 'support': 100}}</td>\n",
       "      <td>0.86</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>Hyperparams-smote-AdaBoostClassifier</td>\n",
       "      <td>0.501836</td>\n",
       "      <td>{'0': {'precision': 0.875, 'recall': 0.9210526315789473, 'f1-score': 0.8974358974358975, 'support': 76}, '1': {'precision': 0.7, 'recall': 0.5833333333333334, 'f1-score': 0.6363636363636365, 'support': 24}, 'accuracy': 0.84, 'macro avg': {'precision': 0.7875, 'recall': 0.7521929824561404, 'f1-score': 0.766899766899767, 'support': 100}, 'weighted avg': {'precision': 0.833, 'recall': 0.84, 'f1-score': 0.8347785547785549, 'support': 100}}</td>\n",
       "      <td>0.84</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>Hyperparams-nonbalance-AdaBoostClassifier</td>\n",
       "      <td>0.243534</td>\n",
       "      <td>{'0': {'precision': 0.8641975308641975, 'recall': 0.9210526315789473, 'f1-score': 0.89171974522293, 'support': 76}, '1': {'precision': 0.6842105263157895, 'recall': 0.5416666666666666, 'f1-score': 0.6046511627906976, 'support': 24}, 'accuracy': 0.83, 'macro avg': {'precision': 0.7742040285899935, 'recall': 0.7313596491228069, 'f1-score': 0.7481854540068138, 'support': 100}, 'weighted avg': {'precision': 0.8210006497725796, 'recall': 0.83, 'f1-score': 0.8228232854391941, 'support': 100}}</td>\n",
       "      <td>0.83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>Hyperparams-smote-XGBClassifier</td>\n",
       "      <td>0.899158</td>\n",
       "      <td>{'0': {'precision': 0.8641975308641975, 'recall': 0.9210526315789473, 'f1-score': 0.89171974522293, 'support': 76}, '1': {'precision': 0.6842105263157895, 'recall': 0.5416666666666666, 'f1-score': 0.6046511627906976, 'support': 24}, 'accuracy': 0.83, 'macro avg': {'precision': 0.7742040285899935, 'recall': 0.7313596491228069, 'f1-score': 0.7481854540068138, 'support': 100}, 'weighted avg': {'precision': 0.8210006497725796, 'recall': 0.83, 'f1-score': 0.8228232854391941, 'support': 100}}</td>\n",
       "      <td>0.83</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>Hyperparams-nonbalance-DecisionTreeClassifier</td>\n",
       "      <td>0.180763</td>\n",
       "      <td>{'0': {'precision': 0.8625, 'recall': 0.9078947368421053, 'f1-score': 0.8846153846153847, 'support': 76}, '1': {'precision': 0.65, 'recall': 0.5416666666666666, 'f1-score': 0.5909090909090908, 'support': 24}, 'accuracy': 0.82, 'macro avg': {'precision': 0.7562500000000001, 'recall': 0.7247807017543859, 'f1-score': 0.7377622377622377, 'support': 100}, 'weighted avg': {'precision': 0.8115000000000001, 'recall': 0.82, 'f1-score': 0.8141258741258741, 'support': 100}}</td>\n",
       "      <td>0.82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>Hyperparams-oversampling-XGBClassifier</td>\n",
       "      <td>0.917828</td>\n",
       "      <td>{'0': {'precision': 0.8717948717948718, 'recall': 0.8947368421052632, 'f1-score': 0.8831168831168831, 'support': 76}, '1': {'precision': 0.6363636363636364, 'recall': 0.5833333333333334, 'f1-score': 0.6086956521739131, 'support': 24}, 'accuracy': 0.82, 'macro avg': {'precision': 0.754079254079254, 'recall': 0.7390350877192983, 'f1-score': 0.745906267645398, 'support': 100}, 'weighted avg': {'precision': 0.8152913752913753, 'recall': 0.82, 'f1-score': 0.8172557876905703, 'support': 100}}</td>\n",
       "      <td>0.82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Production-GradientBoostingClassifier</td>\n",
       "      <td>0.285110</td>\n",
       "      <td>{'0': {'precision': 0.9672131147540983, 'recall': 0.7866666666666666, 'f1-score': 0.8676470588235293, 'support': 75}, '1': {'precision': 0.5897435897435898, 'recall': 0.92, 'f1-score': 0.71875, 'support': 25}, 'accuracy': 0.82, 'macro avg': {'precision': 0.7784783522488441, 'recall': 0.8533333333333333, 'f1-score': 0.7931985294117647, 'support': 100}, 'weighted avg': {'precision': 0.8728457335014713, 'recall': 0.82, 'f1-score': 0.830422794117647, 'support': 100}}</td>\n",
       "      <td>0.82</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>Hyperparams-oversampling-DecisionTreeClassifier</td>\n",
       "      <td>0.323999</td>\n",
       "      <td>{'0': {'precision': 0.9, 'recall': 0.8289473684210527, 'f1-score': 0.8630136986301371, 'support': 76}, '1': {'precision': 0.5666666666666667, 'recall': 0.7083333333333334, 'f1-score': 0.6296296296296297, 'support': 24}, 'accuracy': 0.8, 'macro avg': {'precision': 0.7333333333333334, 'recall': 0.7686403508771931, 'f1-score': 0.7463216641298833, 'support': 100}, 'weighted avg': {'precision': 0.82, 'recall': 0.8, 'f1-score': 0.8070015220700154, 'support': 100}}</td>\n",
       "      <td>0.80</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>Hyperparams-nonbalance-XGBClassifier</td>\n",
       "      <td>0.601304</td>\n",
       "      <td>{'0': {'precision': 0.7551020408163265, 'recall': 0.9736842105263158, 'f1-score': 0.8505747126436782, 'support': 76}, '1': {'precision': 0.0, 'recall': 0.0, 'f1-score': 0.0, 'support': 24}, 'accuracy': 0.74, 'macro avg': {'precision': 0.37755102040816324, 'recall': 0.4868421052631579, 'f1-score': 0.4252873563218391, 'support': 100}, 'weighted avg': {'precision': 0.5738775510204082, 'recall': 0.74, 'f1-score': 0.6464367816091955, 'support': 100}}</td>\n",
       "      <td>0.74</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>Hyperparams-nonbalance-GradientBoostingClassifier</td>\n",
       "      <td>0.292970</td>\n",
       "      <td>{'0': {'precision': 0.7659574468085106, 'recall': 0.9473684210526315, 'f1-score': 0.8470588235294116, 'support': 76}, '1': {'precision': 0.3333333333333333, 'recall': 0.08333333333333333, 'f1-score': 0.13333333333333333, 'support': 24}, 'accuracy': 0.74, 'macro avg': {'precision': 0.549645390070922, 'recall': 0.5153508771929824, 'f1-score': 0.4901960784313725, 'support': 100}, 'weighted avg': {'precision': 0.6621276595744681, 'recall': 0.74, 'f1-score': 0.6757647058823528, 'support': 100}}</td>\n",
       "      <td>0.74</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                             model_name  training_time  \\\n",
       "5          Hyperparams-smote-GradientBoostingClassifier       0.509681   \n",
       "9   Hyperparams-oversampling-GradientBoostingClassifier       0.470805   \n",
       "10          Hyperparams-oversampling-AdaBoostClassifier       0.439040   \n",
       "7              Hyperparams-smote-DecisionTreeClassifier       0.333900   \n",
       "6                  Hyperparams-smote-AdaBoostClassifier       0.501836   \n",
       "2             Hyperparams-nonbalance-AdaBoostClassifier       0.243534   \n",
       "4                       Hyperparams-smote-XGBClassifier       0.899158   \n",
       "3         Hyperparams-nonbalance-DecisionTreeClassifier       0.180763   \n",
       "8                Hyperparams-oversampling-XGBClassifier       0.917828   \n",
       "0                 Production-GradientBoostingClassifier       0.285110   \n",
       "11      Hyperparams-oversampling-DecisionTreeClassifier       0.323999   \n",
       "0                  Hyperparams-nonbalance-XGBClassifier       0.601304   \n",
       "1     Hyperparams-nonbalance-GradientBoostingClassifier       0.292970   \n",
       "\n",
       "                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                       performance  \\\n",
       "5                  {'0': {'precision': 0.9333333333333333, 'recall': 0.9210526315789473, 'f1-score': 0.9271523178807947, 'support': 76}, '1': {'precision': 0.76, 'recall': 0.7916666666666666, 'f1-score': 0.7755102040816326, 'support': 24}, 'accuracy': 0.89, 'macro avg': {'precision': 0.8466666666666667, 'recall': 0.8563596491228069, 'f1-score': 0.8513312609812136, 'support': 100}, 'weighted avg': {'precision': 0.8917333333333335, 'recall': 0.89, 'f1-score': 0.8907582105689958, 'support': 100}}   \n",
       "9       {'0': {'precision': 0.9315068493150684, 'recall': 0.8947368421052632, 'f1-score': 0.912751677852349, 'support': 76}, '1': {'precision': 0.7037037037037037, 'recall': 0.7916666666666666, 'f1-score': 0.7450980392156864, 'support': 24}, 'accuracy': 0.87, 'macro avg': {'precision': 0.8176052765093861, 'recall': 0.8432017543859649, 'f1-score': 0.8289248585340177, 'support': 100}, 'weighted avg': {'precision': 0.8768340943683408, 'recall': 0.87, 'f1-score': 0.87251480457955, 'support': 100}}   \n",
       "10      {'0': {'precision': 0.9315068493150684, 'recall': 0.8947368421052632, 'f1-score': 0.912751677852349, 'support': 76}, '1': {'precision': 0.7037037037037037, 'recall': 0.7916666666666666, 'f1-score': 0.7450980392156864, 'support': 24}, 'accuracy': 0.87, 'macro avg': {'precision': 0.8176052765093861, 'recall': 0.8432017543859649, 'f1-score': 0.8289248585340177, 'support': 100}, 'weighted avg': {'precision': 0.8768340943683408, 'recall': 0.87, 'f1-score': 0.87251480457955, 'support': 100}}   \n",
       "7                   {'0': {'precision': 0.918918918918919, 'recall': 0.8947368421052632, 'f1-score': 0.9066666666666667, 'support': 76}, '1': {'precision': 0.6923076923076923, 'recall': 0.75, 'f1-score': 0.7199999999999999, 'support': 24}, 'accuracy': 0.86, 'macro avg': {'precision': 0.8056133056133057, 'recall': 0.8223684210526316, 'f1-score': 0.8133333333333332, 'support': 100}, 'weighted avg': {'precision': 0.8645322245322246, 'recall': 0.86, 'f1-score': 0.8618666666666667, 'support': 100}}   \n",
       "6                                                          {'0': {'precision': 0.875, 'recall': 0.9210526315789473, 'f1-score': 0.8974358974358975, 'support': 76}, '1': {'precision': 0.7, 'recall': 0.5833333333333334, 'f1-score': 0.6363636363636365, 'support': 24}, 'accuracy': 0.84, 'macro avg': {'precision': 0.7875, 'recall': 0.7521929824561404, 'f1-score': 0.766899766899767, 'support': 100}, 'weighted avg': {'precision': 0.833, 'recall': 0.84, 'f1-score': 0.8347785547785549, 'support': 100}}   \n",
       "2      {'0': {'precision': 0.8641975308641975, 'recall': 0.9210526315789473, 'f1-score': 0.89171974522293, 'support': 76}, '1': {'precision': 0.6842105263157895, 'recall': 0.5416666666666666, 'f1-score': 0.6046511627906976, 'support': 24}, 'accuracy': 0.83, 'macro avg': {'precision': 0.7742040285899935, 'recall': 0.7313596491228069, 'f1-score': 0.7481854540068138, 'support': 100}, 'weighted avg': {'precision': 0.8210006497725796, 'recall': 0.83, 'f1-score': 0.8228232854391941, 'support': 100}}   \n",
       "4      {'0': {'precision': 0.8641975308641975, 'recall': 0.9210526315789473, 'f1-score': 0.89171974522293, 'support': 76}, '1': {'precision': 0.6842105263157895, 'recall': 0.5416666666666666, 'f1-score': 0.6046511627906976, 'support': 24}, 'accuracy': 0.83, 'macro avg': {'precision': 0.7742040285899935, 'recall': 0.7313596491228069, 'f1-score': 0.7481854540068138, 'support': 100}, 'weighted avg': {'precision': 0.8210006497725796, 'recall': 0.83, 'f1-score': 0.8228232854391941, 'support': 100}}   \n",
       "3                              {'0': {'precision': 0.8625, 'recall': 0.9078947368421053, 'f1-score': 0.8846153846153847, 'support': 76}, '1': {'precision': 0.65, 'recall': 0.5416666666666666, 'f1-score': 0.5909090909090908, 'support': 24}, 'accuracy': 0.82, 'macro avg': {'precision': 0.7562500000000001, 'recall': 0.7247807017543859, 'f1-score': 0.7377622377622377, 'support': 100}, 'weighted avg': {'precision': 0.8115000000000001, 'recall': 0.82, 'f1-score': 0.8141258741258741, 'support': 100}}   \n",
       "8      {'0': {'precision': 0.8717948717948718, 'recall': 0.8947368421052632, 'f1-score': 0.8831168831168831, 'support': 76}, '1': {'precision': 0.6363636363636364, 'recall': 0.5833333333333334, 'f1-score': 0.6086956521739131, 'support': 24}, 'accuracy': 0.82, 'macro avg': {'precision': 0.754079254079254, 'recall': 0.7390350877192983, 'f1-score': 0.745906267645398, 'support': 100}, 'weighted avg': {'precision': 0.8152913752913753, 'recall': 0.82, 'f1-score': 0.8172557876905703, 'support': 100}}   \n",
       "0                              {'0': {'precision': 0.9672131147540983, 'recall': 0.7866666666666666, 'f1-score': 0.8676470588235293, 'support': 75}, '1': {'precision': 0.5897435897435898, 'recall': 0.92, 'f1-score': 0.71875, 'support': 25}, 'accuracy': 0.82, 'macro avg': {'precision': 0.7784783522488441, 'recall': 0.8533333333333333, 'f1-score': 0.7931985294117647, 'support': 100}, 'weighted avg': {'precision': 0.8728457335014713, 'recall': 0.82, 'f1-score': 0.830422794117647, 'support': 100}}   \n",
       "11                                  {'0': {'precision': 0.9, 'recall': 0.8289473684210527, 'f1-score': 0.8630136986301371, 'support': 76}, '1': {'precision': 0.5666666666666667, 'recall': 0.7083333333333334, 'f1-score': 0.6296296296296297, 'support': 24}, 'accuracy': 0.8, 'macro avg': {'precision': 0.7333333333333334, 'recall': 0.7686403508771931, 'f1-score': 0.7463216641298833, 'support': 100}, 'weighted avg': {'precision': 0.82, 'recall': 0.8, 'f1-score': 0.8070015220700154, 'support': 100}}   \n",
       "0                                                {'0': {'precision': 0.7551020408163265, 'recall': 0.9736842105263158, 'f1-score': 0.8505747126436782, 'support': 76}, '1': {'precision': 0.0, 'recall': 0.0, 'f1-score': 0.0, 'support': 24}, 'accuracy': 0.74, 'macro avg': {'precision': 0.37755102040816324, 'recall': 0.4868421052631579, 'f1-score': 0.4252873563218391, 'support': 100}, 'weighted avg': {'precision': 0.5738775510204082, 'recall': 0.74, 'f1-score': 0.6464367816091955, 'support': 100}}   \n",
       "1   {'0': {'precision': 0.7659574468085106, 'recall': 0.9473684210526315, 'f1-score': 0.8470588235294116, 'support': 76}, '1': {'precision': 0.3333333333333333, 'recall': 0.08333333333333333, 'f1-score': 0.13333333333333333, 'support': 24}, 'accuracy': 0.74, 'macro avg': {'precision': 0.549645390070922, 'recall': 0.5153508771929824, 'f1-score': 0.4901960784313725, 'support': 100}, 'weighted avg': {'precision': 0.6621276595744681, 'recall': 0.74, 'f1-score': 0.6757647058823528, 'support': 100}}   \n",
       "\n",
       "    f1_score_avg  \n",
       "5           0.89  \n",
       "9           0.87  \n",
       "10          0.87  \n",
       "7           0.86  \n",
       "6           0.84  \n",
       "2           0.83  \n",
       "4           0.83  \n",
       "3           0.82  \n",
       "8           0.82  \n",
       "0           0.82  \n",
       "11          0.80  \n",
       "0           0.74  \n",
       "1           0.74  "
      ]
     },
     "execution_count": 107,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.set_option('display.max_colwidth', None)\n",
    "# pd.reset_option('display.max_colwidth')\n",
    "\n",
    "training_logs_test[[\"model_name\",\"training_time\",\"performance\",\"f1_score_avg\"]].sort_values([\"f1_score_avg\"], ascending=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "GridSearchCV(estimator=GradientBoostingClassifier(), n_jobs=-1,\n",
       "             param_grid={'learning_rate': [0.5], 'max_depth': [2],\n",
       "                         'n_estimators': [5]})"
      ]
     },
     "execution_count": 108,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model_test[\"model_data\"][\"model_object\"]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# 9. Check Production Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [],
   "source": [
    "check_model = util.pickle_load(config_data['production_model_path'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'model_data': {'model_name': 'GradientBoostingClassifier',\n",
       "  'model_object': GridSearchCV(estimator=GradientBoostingClassifier(), n_jobs=-1,\n",
       "               param_grid={'learning_rate': [0.5], 'max_depth': [2],\n",
       "                           'n_estimators': [5]}),\n",
       "  'model_uid': 'fa0af7ba956c1b9675f7f0ffe2f7467d'},\n",
       " 'model_log': {'model_name': 'Production-GradientBoostingClassifier',\n",
       "  'model_uid': 'fa0af7ba956c1b9675f7f0ffe2f7467d',\n",
       "  'training_time': 0.509681,\n",
       "  'training_date': '2023-01-07 22:16:04.572480',\n",
       "  'performance': {'0': {'precision': 0.9333333333333333,\n",
       "    'recall': 0.9210526315789473,\n",
       "    'f1-score': 0.9271523178807947,\n",
       "    'support': 76},\n",
       "   '1': {'precision': 0.76,\n",
       "    'recall': 0.7916666666666666,\n",
       "    'f1-score': 0.7755102040816326,\n",
       "    'support': 24},\n",
       "   'accuracy': 0.89,\n",
       "   'macro avg': {'precision': 0.8466666666666667,\n",
       "    'recall': 0.8563596491228069,\n",
       "    'f1-score': 0.8513312609812136,\n",
       "    'support': 100},\n",
       "   'weighted avg': {'precision': 0.8917333333333335,\n",
       "    'recall': 0.89,\n",
       "    'f1-score': 0.8907582105689958,\n",
       "    'support': 100}},\n",
       "  'f1_score_avg': 0.89,\n",
       "  'data_configurations': 'smote'}}"
      ]
     },
     "execution_count": 110,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "check_model"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.11.3"
  },
  "vscode": {
   "interpreter": {
    "hash": "a90aeebcf29d64a654773811cc170cb25061cb2498f10ac689db374c7bf325de"
   }
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}