Overview

These visualizations show how the model's performance changes under different perturbation conditions.

Performance Analysis

These visualizations show how the model responds to different perturbation levels:

  • Regular Score: Overall performance score across perturbation levels
  • Worst Score: Minimum performance observed at each perturbation level, showing worst-case scenarios
  • Mean Score: Average performance across multiple tests at each level, showing expected behavior

The decline in performance as perturbation level increases indicates the model's sensitivity to feature disturbances.

Perturbation Results

Level Base Score Perturbed Score Impact Worst Score
Level Base Score Perturbed Score Impact Worst Score
{% if has_alternative_models %}

Model Comparison

Compare performance of different models under the same perturbation conditions.

Model Comparison Analysis

Model Base Score Robustness Score Raw Impact Quantile Impact
{% endif %}