We've all had amazing song ideas, but it's hard to actually write a song and see how the idea does, so we built this app using 200+ songs from Spotify and ML to tell you how popular your song will be.
First install flask, numpy, pandas, joblib, and scikit-learn.
Next, run
export FLASK_APP=main.py
export FLASK_ENV=development
python3 rand_forest.py
flask runto run and have flask automatically refresh. Now, go to localhost:5000. The web application should appear.
Note: Before doing flask run, run python3 rand_forest.py. The trained RandomForestRegressor will be dumped on disk as a pickle file with joblib. Once the rand_forest.joblib file is on disk, the code will be able to predict the popularity of user input.
Dataset source: https://www.kaggle.com/zaheenhamidani/ultimate-spotify-tracks-db
Description of Features: https://developer.spotify.com/documentation/web-api/reference/tracks/get-audio-features/