[MADLIB-43] kmeans: additional distance metrics

• Type: Sub-task
• Status: Closed
• Priority: Minor
• Resolution: Unresolved
• Assignee: Aleksander Gorajek
• Reporter: Aleksander Gorajek
• Created: 18/Mar/11 20:42
• Updated: 24/Jan/12 21:50

Description

Add more distance measures to k-means clustering:
http://en.wikipedia.org/wiki/Cluster_analysis#Distance_measure

Use/Create reusable primitive functions.

Mahout has: Euclidean, SquaredEuclidean, Manhattan, Cosine, Tanimoto, WeightedDimensions