Distribution Shift Resilience Analysis

This analysis evaluates how the model's performance changes when data distributions shift. A resilient model maintains consistent performance despite distribution shifts between training and deployment data.

Key Findings

Overall Resilience
{% if resilience_score is not defined or resilience_score is none %} Data not available {% elif resilience_score >= 0.9 %} Excellent ({{ (resilience_score * 100)|int }}%) {% elif resilience_score >= 0.8 %} Good ({{ (resilience_score * 100)|int }}%) {% elif resilience_score >= 0.7 %} Moderate ({{ (resilience_score * 100)|int }}%) {% elif resilience_score >= 0.6 %} Fair ({{ (resilience_score * 100)|int }}%) {% else %} Limited ({{ (resilience_score * 100)|int }}%) {% endif %}
Performance Gap
{% if performance_gap is not defined or performance_gap is none %} Data not available {% elif performance_gap < 0.05 %} Minimal ({{ (performance_gap * 100)|round(1) }}%) {% elif performance_gap < 0.1 %} Small ({{ (performance_gap * 100)|round(1) }}%) {% elif performance_gap < 0.15 %} Moderate ({{ (performance_gap * 100)|round(1) }}%) {% elif performance_gap < 0.25 %} Significant ({{ (performance_gap * 100)|round(1) }}%) {% else %} Large ({{ (performance_gap * 100)|round(1) }}%) {% endif %}
Feature Sensitivity
{% if sensitive_features is defined %}{{ sensitive_features|length }}{% else %}0{% endif %} highly sensitive features
Most Affected Scenario
{% if most_affected_scenario is defined and max_gap is defined %} {{ most_affected_scenario }} ({{ (max_gap * 100)|round(1) }}% gap) {% else %} Data not available {% endif %}

Recommendations