Examples

See examples here: https://github.com/MobileTeleSystems/RecTools/tree/main/examples

• Simple example of building recommendations with RecTools
  • Load data
  • Build model
  • Check recommendations
• Example of model selection using cross-validation with RecTools
  • Load data
  • Prepare cross-validation splitter
  • Train models
• Examples of calculating different metrics with RecTools
  • We provide all types of metrics to measure model performance from different aspects
    • Load and preprocess data: Movielens
    • Build model
    • Calculate metrics
      • Metrics initialization
  • Simple metrics
  • Metric with complex additional parameter
    • Single metric calculation
      • Per user metric calculation
    • Multiple metrics calculation with one function
• Examples of constructing datasets with features in RecTools
  • Load data: Movielens 1m
  • Data types: categorical and numerical
  • Features storage: Sparse example
  • Features storage: Dense example
  • Feeding features to models
  • Final notes
• Benchmark quality: RecTools wrapper for iALS with features
  • Load and preprocess data: Kion
  • Datasets
    • Metrics
    • Model initialization
  • Grid Search for best pure iALS model params
  • Add different feature selections to 3 best iALS models
  • Visualise all results
  • Summary on our benchmark
  • Technical details on RecTools implementation of iALS with features
• Benchmark inference speed: RecTools wrapper for LightFM
  • Load data: Movielens 20m
  • Train LightFM model with different latent factors dimension size
  • Compute inference speed for both frameworks on all models
  • Benchmark frameworks
  • Compare the difference
    • Inference speed
    • Other benefits of RecTools wrapper
    • How is the speed optimized?
• Example of VisualApp for recommendations
  • Prepare data
  • Simple visualization example
  • Details on VisualApp features
    • 1. How to process pd.DataFrame values into any desired html code (e.g. display actual images from links)
    • 2. How to add random users for visualization
    • 3. How to pass multiple models for visualization
    • 4. How to save VisualApp and easily access the same widgets later
    • 5. Hot to visualize item-to-item recommendations
  • Final app example