For those interested, I designed a black-box, automated (and very simple) computation of the optimum number of clusters, based on the elbow method: https://www.datasciencecentral.com/profiles/blogs/how-to-automatically-determine-the-number-of-clusters-in-your-dat