Echo Nest Taste Profile User Data Processor

Python script to create a sparse matrix (scipy csr matrix) of the user data provided by Echo Nest.
Resulting matrix is song x user matrix with play count as the data.

Data format

• User data is provided by Echo Nest in a tab delimited .txt file. Each row contains user_id, song_id and play_count.

To use

• Download user data (train_triplets.txt) from MSD website and change base_dir to the path where the file is saved
• If subset of the data is used (ex 20000 users, 10000 songs)
  python load_user_data.py --users 20000 --songs 10000
• If all data is used
  python load_user_data.py
• After running the script file, three files will be created in the same directory as the script file :
  • song_user_matrix_20000_10000.npz : sparse song x user matrix
  • subset_songs_20000_10000.npy : numpy array of songs in the sparse matrix (order preserved)
  • subset_users_20000_10000.npy% : numpy array of users in the sparset matrix (order preserved)
• To load user matrix
  song_user_matrix = scipy.sparse.load_npz('song_user_matrix_20000_10000.npz')

Echo Nest Song ID to MSD Track ID

The user data is given with the Echo Nest song id.
In case you want to (of course, you need the MSD audio files)process audio data, you need to convert the song id to MSD track id.

File below provides the mapping between echo nest id and msd id :
echonest_id_to_msd_id.pkl.

Adapted from https://github.com/jongpillee/music_dataset_split/tree/master/MSD_split, so check this out for usage.