Sentiment Analysis with Logistic Regression
Overview
This is the code for this video on Youtube by Siraj Raval.
This repository contains a jupyter notebook and the necessary data to implement sentiment analysis of tweets using Logistic Regression. Please open the notebook for more information.
The dataset
The dataset was obtained from a Kaggle competition. The dataset is divided into a train and a test dataset. Each record contains the following fields:
Field name | Meaning |
---|---|
ItemID | id of twit |
Sentiment | sentiment (1-positive, 0-negative) |
SentimentText | text of the twit |
Web app
You can go straight ahead and try out the algorithm with a small web app I have included in this repository, just run:
cd site
python app.py
Then open a browser in the default address (http://127.0.0.1:5000/
) and play around:
Requirements
This notebook will run in Python >= 3.5. The following packages are required:
- bokeh
- flask
- nltk
- numpy
- pandas
- scikit-learn
Limitations
Because the training set contains only English twits, this classifier will only work with English twits.
Credits
Credits for this code go to guillermo-carrasco