K-Means and Cluster Metrics

K-Means and Cluster Metrics

Cluster Metrics

K-Means is probably the first algorithm that comes to mind when someone mentions unsupervised learning or clustering.

One of the downsides of k-Means is that you need to determine how many clusters there should be. One way to do this is to run k-Mean multiple times, with different numbers of clusters, and use one of several metrics to determine which is the best clustering.

This post goes through 4 different techniques / metrics for determining the number of clusters.

Again, this is one of just a few posts on this site that is authored by myself.