Usage: wkmeans [options] < input > output -k number of clusters (default 2) -r number of restarts (default 0) -s random seed -l input file contains labels -w input file contains instance weights -v verbose output

Input format (assuming you have m vectors of n dimensions): [label_1] [weight_1] x_11 ... x_1m ... [label_m] [weight_m] x_m1 ... x_mn

label_i : (string) label of the ith vector, required when -l used weight_i : (double) weight of the ith vector, required when -w used x_ij : (double) ith vector's jth component

Output format: [label_1] c_1 ... [label_m] c_m

c_i : (int) cluster of ith vector

Algorithm: wkmeans is a k-means algorithm with (optional) instance weights.

Based on mpi_kmeans-1.5 by Peter Gehler.
Based on C. Elkan. Using the triangle inequality to accelerate kMeans. ICML 2003.
Initialization based on Arthur, D. and Vassilvitskii, S. (2007). K-means++: the advantages of careful seeding. Proceedings of the eighteenth annual ACM-SIAM symposium on Discrete algorithms. pp. 1027-1035.

Please see the file LICENSE for terms of use. Everything is standard C, so just typing make should give you an executable.

Wkmeans

Install / Use

README