Hierarchical runs too slow

[cheesebear]

Hierarchical seems to run np.min too many times. I suggest apply stack() to dists and sort the dataframe first, to avoid running min repeatedly.

[wannesm]

You are correct that, this method can be further optimized and the usage of min is not ideal. Do you have a worked out solution? I assume you imply argsort instead of sort?

For fast clustering implementations I recommend to use the Scipy and Pyclustering implementations. They are already wrapped in clustering.hierarchical.LinkageTree and clustering.medoids.Medoids, respectively. And by looking at the code in those classes, one can use almost all other clustering techniques in the other toolbox.