YouTube Comments Spam Classifier

YouTube Walkthrough

• Training the model
• Deploying Flask API

Running locally to generate predictions

Run the Flask API

• install pipenv
• run python server.py
• send requests to POST localhost:5000/predict or use the Swagger UI to generate predictions

OR, run the Jupyter notebook

Limitations

• dataset only comes from comments on music videos of five artists so the model may not perform as expected in other contexts (e.g. a tech video)
• what is considered spam in one context may just as well be considered non-spam in a different context