On: Stop using the elbow criterion for k-means

[pbronez]

Paper points out that K-means is a special case of gaussian mixture modeling that makes lots of assumptions about the data. As always, understanding the statistical assumptions behind your model is absolutely essential.