Spotify Recsys 2018 challenge submission for TeamFL
Switch branches/tags
Nothing to show
Clone or download
Latest commit 81e00a2 Jul 2, 2018
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
LICENSE Update LICENSE Jul 2, 2018
README.md Update README.md Jul 2, 2018
clean_prediction.py Add files via upload Jul 2, 2018
mpd_dataset_generator.py Add files via upload Jul 2, 2018
mpd_slicer.py Add files via upload Jul 2, 2018
submission.py Add files via upload Jul 2, 2018
word2vec.py Add files via upload Jul 2, 2018
word2vec_prediction.py Add files via upload Jul 2, 2018

README.md

Recsys_challenge_2018_TeamFL

How to run the codes:

Step 1: Run mpd_slicer.py on the mpd dataset to generate 1000 different json files.You need to create a folder named 1000_playlists before running the code. The 1000 different json files will be saved in that folder: python mpd_slicer.py

Step 2: Run mpd_dataset_generator.py to pull relevant information from the mpd for training. This code takes one command line argument which is the path to the folder where the 1000 json files are kept: python mpd_dataset_generator.py 1000_playlists

Step 3: Run word2vec.py to train the model on the generated mpd_dataset: python word2vec.py

Step 4: Run challenge_slicer.py on the challenge dataset to generate 1 json file. You need to create a folder named 1_playlist before running the code. The json file will be saved in that folder: python challenge_slicer.py

Step 5: Run challenge_dataset_generator.py to pull relevant information from the challenge dataset for prediction. This code takes one command line argument which is the path to the folder where the json file is kept: python challenge_dataset_generator.py 1_playlist

Step 6: Run word2vec_prediction.py to generate the prediction.

Step 7: Run clean_prediction.py to clean up the prediction.

Step8: Run submission.py to get the 500 tracks for each prediction.