inst/helpfiles/measures_help.md

In rhenkin/visxhclust: A Shiny App for Visual Exploration of Hierarchical Clustering

Internal validation

This tab show the results of a selected internal validation measures for k clusters ranging from 2 to 14. The clusterCrit package is used to compute the measures/indices. For more information, please check the documentation of the package at [https://cran.r-project.org/web/packages/clusterCrit/index.html]. To facilitate interpretation, it is possible to highlight first or global minimum or maximum values.

The following is a short summary about which values to look for in each metric: Maximum: Calinski-Harabasz, Dunn, GDI, Gamma, PBM, Point biserial, Ratwosky-Lance, Silhouette, Tau, Wenmert-Gancarski Minimum: Banfeld-Raftery, C-index, Davies-Bouldin, G-plus, McClain-Rao, Ray-Turi, Scott-Symons, SD, Xie-Beni

The remaining measures are the maximum or minimum difference between consecutive k's: Max diff: Ball-Hall, Ksq, Trace Min diff: Det ratio, log Det, log SS

rhenkin/visxhclust documentation built on March 28, 2023, 8:16 p.m.