Feature Model Importance Robustness Importance Model Rank Robustness Rank Rank Difference In Optimal Subset
Loading feature comparison data...
Page 1 of 1

Importance Comparison Analysis

This table compares model-derived importance with robustness-based importance. Features with large rank differences may be particularly interesting:

  • High model importance, low robustness importance: Important for predictions but stable under perturbation
  • Low model importance, high robustness importance: Less important for predictions but highly sensitive to perturbation

Features in the optimal subset (highlighted) represent the best balance between predictive power and robustness.