This is a simple ML model that is trained to predict the sentiment depicted by a tweet.
A private dataset was used for training was taken to train the model.
The model was trained in Python using four modules: sklearn, nltk, re and pandas.
Python file sentiment_analysis.py contains all the clean code to be used, although jupyter notebook (unclean code) is also present.
Model was trained using sklearn pipeline: CountVectorizer, TfidfTransformer and LinearSVC classifier.
The training data contained only 2 labels - Positive and Negative Sentiment but test data also had Neutral Sentiment.
Thats why accuracy score was better for train set (79.33) and validation set (76.50) but less for test set (60.44).
Run the sentiment_analysis.py to train the model and Evaluate sentiment for any tweet.
If you like this project, please give this repo a ⭐.