A simple clustering evaluation of KMeans for the Weka

Silhouette-Index in IRIS dataset

KValid is a simple clustering evaluation package for WEKA. It uses the SimpleKMeans algorithm as a backend to cluster the instances and evaluates the clusterer using some algorithms, currently Silhouette-Index and Elbow.

KValid uses Silhouette-Index and Elbow to validate the SimpleKMeans algorithm. Besides calculating the SI and SSE, the package tell which is the best K and allows plot the graph into the screen and save as PNG format.

In order to install KValid, download trough the release menu in GitHub, this link to be more specific. (make sure that your WEKA version is >= 3.7.1)

Once downloaded, install through the package manager: Tools -> Package manager -> File/URL button -> Browse (searches for the KValid.zip file) and click OK.

A confirmation message like 'Weka will need to be restarted after installation...", will appears. Just close and open Weka again and the package should works.

Since KValid is a cluster evaluator, you can find it in the Cluster menu.