Skmeans

The datasets are uploaded in the mydata folder. The last column denotes the ground truth. The R code for the S-k-means algorithm and other peer algorithms are uploaded in the functions.R file. An example run is given in Example.pdf and Example.Rmd.

Details

The details for S-k-means are as follows:

Inputs:

1. X : An n * p matrix to be clustered, whose rows denote the data points.
2. M : A k * p matrix, whose rows denote the inital cluster centroids.
3. itermax : Maximum number of itearions to run. Default is 30.

Outputs:

1. label : An n-length vector denoting the class labels.
2. centroids : A k * p matrix, whose rows denote the cluster centroids.

Paper

S. Chakraborty, S. Das, k−Means clustering with a new divergence-based distance metric: Convergence and performance analysis, Pattern Recognition Letters (2017), https://doi.org/10.1016/j.patrec.2017.09.025