view article
Figure 3
(a) A 3D scatter plot illustrating the k-means clustering results, with k = 2 (Hartigan–Wong algorithm), based on MPEfinal (x axis), Rf (y axis) and CORRfinal (z axis). Data points are colour-coded to distinguish between `Solved structures' (blue) and `Unsolved structures' (red), with the legend indicating the number of observations in each category. (b) A silhouette plot for k-means clustering with k = 2. Each block contains a number of vertical bars corresponding to the elements assigned to that cluster (blue for solved and red for unsolved structures). The height of each bar represents the silhouette width, indicating how well each element fits within its assigned cluster compared with the other one. Silhouette values closer to 1 correspond to better-defined clustering. The value of the mean silhouette width, shown above the plot, quantifies the overall clustering quality; values above 0.7 are generally considered indicative of strong and well separated clusters (Kaufman & Rousseeuw, 1990View full citation).

Journal logoJOURNAL OF
APPLIED
CRYSTALLOGRAPHY
ISSN: 1600-5767
Volume 58| Part 6| December 2025| Pages 1859-1869
https://doi.org/10.1107/S1600576725008271
Follow J. Appl. Cryst.
Sign up for e-alerts
E-alerts
Follow J. Appl. Cryst. on Twitter
Twitter
Follow us on facebook
Facebook
Sign up for RSS feeds
RSS