Support the Underground: Characteristics of Beyond-Mainstream Music Listeners

This reposity includes python scripts and ipython-notebooks necessary for conducting analyses on the LFM-BeyMS dataset, available via Zenodo https://doi.org/10.5281/zenodo.3784765. For more details, we refer to our publication in https://arxiv.org/abs/2102.12188.

Files

To reproduce our results, the python scripts and ipython-notebooks must be executed in following order:

  1. Identification of BeyMS and MS.ipynb: Identifies BeyMS and MS based on mainstreaminess.
  2. Identification and Analysis of Track Clusters.ipynb: Clustering and analysis of tracks listened by BeyMS. Additional statistics on track clusters.
  3. Identification and Analysis of User groups.ipynb: Assign users in BeyMS to track clusters. Additional statistics of user groups.
  4. Rating Dataset Generation.ipynb: Create dataset used for recommendation experiments. Includes both, BeyMS and MS.
  5. Recommendations.py: Run several recommendation algorithms and evaluate them groupwise.
  6. Visualization of Recommendation Performance.ipynb: Visualize results of recommendation experiments.

Requirements

  • Python 3
  • numpy
  • matplotlib
  • pandas
  • seaborn
  • ast
  • sklearn
  • scipy
  • pycountry
  • umap
  • hdbscan
  • surprise
  • statsmodels

Contributors

  • Peter Müllner, Know-Center GmbH, pmuellner [AT] know [MINUS] center [DOT] at (Contact)
  • Dominik Kowald, Know-Center GmBH
  • Markus Schedl, JKU Linz
  • Christine Bauer, JKU Linz
  • Eva Zangerle, University of Innsbruck
  • Elisabeth Lex, Graz University of Technology