AmirInt/sounds_sus

Linear classifier on sentiment analysis. As part of the MITx course on linear models and deep learning.

sounds_sus

This project implements a linear classifier for sentimental analysis purposes on product reviews.

Implementation

Algorithms

The linear classifier takes advantage of:

• The perceptron algorithm
• The average perceptron algorithm
• The Pegasos algorithm

Data Extraction

Use the use_bigram in main.py to set the language model as bigram (True) or unigram (False). Set the remove_stopwords variable to choose to remove the predefined stopwords (e.g. i, me, my, myself, etc.) from the extracted dictionary and the generated feature matrices.

Tests

Run the test.py python script for basic validation tests.

Run

Run the main.py python script to see the results.