Examples

Many of the examples can be viewed (and run) as Jupyter notebooks in the examples directory of the LightFM repository.

• Movielens implicit feedback recommender
  • Implicit feedback
  • Getting the data
  • Fitting models
• Learning rate schedules
  • Preliminaries
  • Experiment
• Cold-start hybrid recommender
  • A pure collaborative filtering model
  • A hybrid model
  • Bonus: tag embeddings
• Learning-to-rank using WARP loss
  • Preliminaries
  • Accuracy
  • Fitting speed
• Building datasets
  • Getting the data
  • Building the ID mappings
  • Building the interactions matrix
  • Building a model