analysing last.fm and spotify listening history. using playlists as models for classification using SVMs and MLPs
Go to file
2021-05-06 16:19:44 +01:00
analysis fixed query numbers, quick cache for playlists 2021-02-03 16:08:06 +00:00
docs added stratifying dataset, added hours per day graph 2021-02-20 00:16:03 +00:00
.gitignore added neural net notebook 2021-05-06 16:19:44 +01:00
album.ipynb added neural net notebook 2021-05-06 16:19:44 +01:00
analysis.ipynb added to playlist cross-validation 2021-03-10 20:01:13 +00:00
artist.ipynb added to playlist cross-validation 2021-03-10 20:01:13 +00:00
playlist-nn.ipynb added neural net notebook 2021-05-06 16:19:44 +01:00
playlist-svm.ipynb added neural net notebook 2021-05-06 16:19:44 +01:00
playlist.ipynb added stratifying dataset, added hours per day graph 2021-02-20 00:16:03 +00:00
poetry.lock added neural net notebook 2021-05-06 16:19:44 +01:00
prep-audio-features.py restructured, added notebooks, refreshed data, poetry 2021-02-01 01:37:22 +00:00
prep-scrobbles.py restructured, added notebooks, refreshed data, poetry 2021-02-01 01:37:22 +00:00
pyproject.toml added neural net notebook 2021-05-06 16:19:44 +01:00
README.md added neural net notebook 2021-05-06 16:19:44 +01:00
stats.ipynb restructured, added notebooks, refreshed data, poetry 2021-02-01 01:37:22 +00:00

Listening Analysis

Notebooks:

  • analysis for a intro to the dataset and premise
  • artist, album & playlist investigations
  • stats for high-level stats about the dataset (Spotify feature miss ratio)
  • playlist SVM using Scikit to classify tracks using the contents of playlists as models
  • playlist NN using a multi-layer perceptron to classify tracks using the contents of playlists as models

Combining Spotify & Last.fm data for exploring habits and trends Uses two data sources,

  1. Last.fm scrobbles
  2. Spotify audio features

The two are joined by searching Last.fm tracks on Spotify to get a Uri, the track name and artist name are provided for the query. These Uris can be used to retrieve Spotify feature descriptors.

SVM Comparison