mczerny

approximate_predict generates new clusters when using cluster_selection_epsilon

8

I am clustering geolocation data, and I noticed that when I try to predict clusters on new data I get more clusters than I saw in the original training data....