Figure 11
From: Finding reproducible cluster partitions for the k-means algorithm
Performance chart for 500 dataset. Ordered Cramérs' V for 100 runs of the SeCo framework with selection of the best solution being (a) SSQ only and (b) Median Internal Cramérs' V for the top 10% of solutions. Results for the dataset with 500 data points. Red lines highlights k = 10, blue lines k = 9 and the black lines k = 8. Confidence intervals for k = 8 (Upper: +0.023, Lower: -0.043) k = 9 (Upper: +0.022, Lower: -0.044) k = 10 (Upper: +0.020, Lower: = -0.042).