CS 294 - 1 Assignment 2 Report

In this report, we describe our implementation of a linear regression method to classify a numerically-scored sentiment data. The dataset was collected by Mark Dredze and others at Johns Hopkins, which records 1M amazon.com book review. The linear regression classification starts with reading tokenized data and building word counts map, and then training linear classifier by minimizing error term. We conduct a 10-fold cross validation to evaluate our method. We will describe our implementation of each of these steps in addition to enhancements such as stop words removing.