The following project takes the tweets from the Twitter Emotion Classification Datase and trains a classifier model that labels a tweet in one of the following 5 categories
| Index | Emotion |
|---|---|
| 0 | sadness |
| 1 | joy |
| 2 | love |
| 3 | anger |
| 4 | fear |
| 5 | surprise |
We evaluated with the following classification models
- Logistic Regression Classifier
- Multinomial Naive Bayes Classifier
- Extra Trees Classifier
Simple Next.js project to use as new Tweet input
To run the frontend, inside /frontend folder, run:
npm run dev
Then go to http://localhost:3000/
The backend uses the trained model along with the vectorizer to classify the Tweet.
To run the backend, inside /backend folder, run:
fastapi dev main.py
Then go to http://127.0.0.1:8000/docs
You can also run the project using docker-compose. Fron root folder run:
docker-compose build
Then run
docker-compose up