Sentiment Analysis on mobile phone reviews

Sentiment Analysis by Hitesh Vaidya

Overview

This repository contains code for sentiment analysis on a dataset of mobile reviews. The dataset is downloaded from Kaggle. The code is developed using Scikit learn. It uses following algorithms:

Bag of Words
Multinomial Naive Bayes
Logistic Regression
Support Vector Machine
Random Forest
Decision Tree

Also, it has visualisation of data and the knowledge obtained from it.

Dependencies

python2.7
virtualenv
jupyter notebook
pandas
numpy
nltk
matplotlib
sklearn
beautifulsoup4
re

Data

Download the required data from this kaggle page.

Installation

You may run this code in a virtual environment. I preferred to do so.
Assuming that you have installed pip and virtualenv,
Create a virtualenv and activate it. eg. let's call it senti

cd senti
git clone https://github.com/hiteshvaidya/sentiment_analysis.git
cd sentiment_analysis
pip -r install requirements.txt

In order to run jupyter notebook in a virtualenv, you need to create a new kernel. Follow this blog or this stackoverflow page to create one.

Usage

Open jupter notebook. Now go in settings and change kernel to the new kernel that you just created.
Now run the code in jupyter notebook.

Acknowledgement

Credits for part of this code to Suki Lau.