A New Measure for Clustering Model Selection

Typical k-means clustering procedures require a priori knowledge of the number of clusters in the data set. This value can be very difficult to ascertain. Existing heuristic methods work in some cases, but are rarely very reliable. Herein, a new method for determining the number of k-means clusters in a given data set is presented. The algorithm is developed from its theoretical basis and its implementation is examined and compared to existing solutions.