Figure 4
From: Finding reproducible cluster partitions for the k-means algorithm
Separation Concordance map for synthetic dataset using 10% threshold for separation metric. Separation-Concordance map for the synthetic data, highlighting the top 10% ΔSSQ on the y-axis and the internal median Cramérs' V on the x-axis. For 500 initialisations of the k-means algorithm, highlighting 50 for each value of k.