(simple) Tool for estimating the number of clusters

12 validity indices, illustrate estimation of the number of clusters

Description

For validation of clustering results in cluster analysis, it is important to use some objective measures to evaluate the clustering quality. This tool provides programs of such validity indices, including 4 External validity indices and 8 Internal validity indices: Rand index, Adjusted Rand index, Silhouette, Calinski-Harabasz, Davies-Bouldin, Homogeneity, Separation, and etc.

This tool is suitable for the performance comparison of different indices on the estimation of the number of clusters, algorithm design for applications by using or improving part codes, and etc.

• Required Products: Statistics Toolbox
• MATLAB release: MATLAB 7.2 (R2006a)

[Günther Eibl]

Documentation of functions (input-output) could be improved, but due to the brevity of the code this does not hurt too much.

[Erik]

It took me a long time to find this. I needed a piece of software to do the Calinski Harabasz method which I was searching for under the name 'Psuedo F'. This is exactly what I needed.