Figure 1: Sample clusters (A-F) Simulated data created by supplying cartesian coordinates for six centroids and generating random coordinates normally distributed around each centroid with sample sizes i) n= 30, 50, 500, 50, 70, 300), and ii) n = 20, 100, 500, 20, 100, 500) with standard deviation =1. The boundaries of each cluster are approximated by circles, colours indicate cluster membership as determined by k-means operating on a matrix of Euclidean distances.