Fig. 3

Comparison of similarity matrices from a dataset for the Small N, large P scenario. In each matrix, the (i, j)th entry is the proportion of clusterings for which the ith and jth items co-clustered for the method in question. In the first row the PSM of the pooled Bayesian samples is compared to the CM for CC(100, 50), with a common ordering of rows and columns in both heatmaps. In the following rows, 6 of the long chains that passed the tests of convergence are shown