This is the individual project of Sahan Bulathwela in partial fullfillment of degree, MSc. Computational Statistics and Machine Learning.

Project Research Area: Topic Classification from user feedback data Plausible Extensions : Sentiment Analysis

Supervisors : Prof. John Shawe-Taylor: CS, UCL Dr. Martin Goodson, Qubit Digital, London

Description:

We attempt to use bag of words features with morphological pre-processing techniques such as spell correction, Stemming to investigate the improvements of performance in Support Vector Machines. We would also like to adapt this algorithm to portray online algorithm nature to improve scalability and performance.

Technologies used: Python scikit-learn Python numpy, scipy, matplotlib, pandas Python nltk