Added divisive hierarchical clustering algorithm that uses sse as selection criteria and euclidean distance as the split measure. The distances between two clusters are found using 'median' (distance between the average of two centroids from another centroid). It works on data having redundant examples as well. #features in the dataset were taken as 2. T.C.: O(#clusters)
A vectorized version of kmeans was also implemented that, I think, runs in O(#clusters).
Other comments
Some improvements can be done like making the code in utils.allocate better, introducing different selection, split measures and distance criteria, etc. I have added notable references in the utils file.
Fixes #94
Brief description of what is fixed or changed
Other comments
Some improvements can be done like making the code in utils.allocate better, introducing different selection, split measures and distance criteria, etc. I have added notable references in the utils file.