Skip to content

ChubingZeng/Sentiment-Analysis

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

7 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Sentiment Analysis for 400,000 Amazon Reviews

Description

In this project, the goal is to perform sentiment analysis to determine whether a review is positive or negative. I implemented 3 different machine learning algorithms to build text classifiers for Amazon reviews. The three algorithms are: neural networks (LSTM to be specific), decision tree and Naive Bayes.

Dataset

The data I'm using comes from the Kaggle Amazon review competition.

Analysis Result

The LSTM model performs the best (AUC 0.96) but took the longest to train.

roc_lstm

Please refer to the .py files for my code, and analysis report.pdf for detailed description of how I pre-processed the data, built up the models and compared the performance of the three methods.

In this 5-min video, I described in detail the dataset, preprocesssing, classifications, results and discussion of the problem.