Machine Learning

This page contains some library code that can be used for machine learning tasks. The focus at this point is learning on text based input.

The following has been implemented:

Distance metrics - Euclidean, Pearson co-relation coefficient
Generic array and file operations useful in handling input data
A library to prepare input text and convert it into a word matrix, on which different algorithms can be applied
An agglomerative hierarchical clustering algorithm

Download or check out from:

http://208.78.102.143/svn/ml/trunk/

A clustering example from sample Wikipedia text has been implemented. The input data is a set of nine files each containing text about one of three selected topics (Biology, Operating Systems, Computer memory). The input files and corresponding clustered output is available here.

Planned

Other distance metrics - Manhattan distance, Jaccard co-efficient
Identifying similarity - Recommendations via collaborative filtering
More unsupervised learning algorithms like k-means and column clustering
Start on supervised learning algorithms

Bugs, suggestions or feedback? Please send it to thejo@kote.in

Last updated: August, 2008

Home