User guide

TODO: define the target audience of the user guide.

Table of contents

• 1. Definition of concepts
  • 1.1. Variable Importance
  • 1.2. Variable Selection
  • 1.3. Variable Selection vs Variable Importance
  • 1.4. Types of VI methods
  • 1.5. High-dimension and correlation
  • 1.6. Statistical Inference
  • 1.7. References
• 2. GLM methods
  • 2.1. Desparsified Lasso
  • 2.2. Knockoffs
• 3. Model-agnostic methods
  • 3.1. Total Sobol Index
  • 3.2. Leave-One-Covariate-Out
  • 3.3. Conditional Feature Importance
    • 3.3.1. Theoretical index
    • 3.3.2. Estimation procedure
    • 3.3.3. Inference
    • 3.3.4. Regression example
    • 3.3.5. Classification example
    • 3.3.6. References
  • 3.4. Permutation Feature Importance
• 4. Marginal methods
  • 4.1. Leave-One-Covariate-In
• 5. Tools for visualization
• 6. Measuring the importance of feature groups
• 7. Inference in high dimension