Python Programming for Biology: Bioinformatics and Beyond

Figure 23.5.  Example output from the k-means clustering method, for both sparse

and dense data. By using an iterative algorithm, the k-means clustering method partitions

data points into a specified number of clusters (k). The cluster centres are initially chosen

at random, and the membership of each data point is determined by its closest cluster

centre. Subsequent cluster centres are defined as the geometric mean of the cluster

members. The cluster assignment and centre recalculation steps are repeated until

convergence. Because membership is based on distance to the cluster centres, the

boundaries between clusters will be straight lines representing the points of equal distance

between two centres.