These tools and metrics are designed to help AI actors develop and use trustworthy AI systems and applications that respect human rights and are fair, transparent, explainable, robust, secure and safe.
Partial Dependence Complexity (PDC)
The Partial Dependence Complexity metric uses the concept of Partial Dependence curve to evaluate how simple this curve can be represented. The partial dependence curve is used to show model predictions are affected on average by each feature. Curves represented by linear functions are easy to grasp and interpret, but some features may have a highly non-linear and complex effects on the model outcomes. Our proposed metric aims to measure how much the curve changes across the variable domain.
Please refer to the reference website to access the full formula.
Trustworthy AI Relevance
This metric addresses Explainability and Robustness by quantifying relevant system properties. Explainability: PDC directly measures the complexity of partial dependence curves (the marginal effect of features on predictions). Low-complexity partial dependence (simple monotonic or linear trends) yields more straightforward, human-understandable explanations; high PDC (many modes, oscillations, sharp changes) signals that single-feature explanations will be hard to interpret and that explanations based on marginal effects may be misleading.
About the metric
You can click on the links to see the associated metrics
Objective(s):
Lifecycle stage(s):
Target users:
