Spotify Recommandation

Create your own recommandation system

This tool allows you to build your own recommandation system based on a few playlists

Features

• Create a recommandation system from a nice-looking GUI made with Gradio
• Build a dataset with audio features of your selected songs
• Build a stat report based on what you listen
• Find your affinity with a specific track, or a list of tracks within a playlist or an album

Installation & Usage

Dillinger requires Python3 and Pip to run.

Install the repository

git clone https://github.com/Brice-Vergnou/create-your-spotify-recomendation.git

Install the dependencies

cd create-your-spotify-recomendation
pip install -r requirements.txt

Create your model ( instructions are going to be given )

python3 get_your_model_gui.py                       # This is going to open your web browser

Use your model ( usage is pretty straight forward for this one )

python3 main.py                                     # This is going to open your web browser

Files

File	Usefulness
data/bad[number].json	Tracks' audio features from a disliked playlist
data/good.json	Tracks' audio features from the liked playlist
data/data.csv	Dataframe made from merging all the data files and adding their "liking state" ( 1 or 0 )
data/model.sav	Saved model
stats/heatmap.png	Correlation map
stats/stats.pdf	Correlation map + explainations about it
flagged/log.csv	If you flag a result in main.py, they'll be stored in this file