Skip to main content
. 2019 May 28;20:266. doi: 10.1186/s12859-019-2889-z

Fig. 4.

Fig. 4

Dunn index, Silhouette width and Davies–Bouldin index for assessing the number of groups in k-means algorithm. The figure presents the evaluation of the number of groups by means of two metrics, which combine the measures of compactness and separation of the clusters. Briefly, the higher the value of both indexes, the better the division into clusters. The figure shows that for the data set examined in the presented work, the optimal number of groups is 4, which agrees with the figure of sensitivity and precision - for example, precision for CODEX tool