Question

I have a distance matrix composed of pair-wise levenshtein’s distance. I was using scikits-learn. But hierarchical clustering algorithm doesn’t take distance matrix as input for clustering. SO I have to search for a new package which can do this.

Are there any fast and well tested packages that you have used for hierarchical clustering ?

Answer 1

The scipy.cluster.hierarchy.ward function (used as backing implementation of sklearn‘s Ward when the connectivity constraints are not required) can accept both observations and precomputed distance (see the online documentation for more details). It’s just not yet exposed in sklearn‘s API.