We setout to build a content based recommender system that could act as your own personal DJ.

For information on how we created the recommender through data analysis, visualization, and training models, take a look at models.ipynb. To explore how we created the DJ, take a look at discjockey.py. For all the data we used to train and test our recommender, look at the musicdata folder. To create your own DJ, take a look at and run main.py.

To use:

1. Create a conda environment using requirements.txt file on your computer and fork this repo.
2. Run main.py.
3. Get your party playlist id and give it to the DJ. Tog get a playlist id, navigate to your desired party playlist on spotify and copying the last part of the url. For example my party playlist id looks like: 4RdNpG06Gzma4AuvfES6QR?si=PhZD_sK5T5C2kxkhQH9-lQ

The DJ will learn from you playlist and only accept songs that are similar to your party preferences.

4. Test our your DJ by passing it a song track! To get a track id, navigate to a song of your choice on spotify, then go to share --> Copy URI. The URI should look something like this: spotify:track:2374M0fQpWi3dLnB54qaLX. Copy the last part after URI after track: so your track id should look something like this: 2374M0fQpWi3dLnB54qaLX.

After inputing your party preferences and track, the DJ will tell you if he accepts your song choice or not!

Hack The Northeast Hackathon submission