Build a comprehensive model evaluation framework with multi-metric assessment, statistical significance testing, and slice-based performance analysis.
## CONTEXT A model that looks great on aggregate metrics can hide catastrophic failures on critical subpopulations — a loan approval model with 92% overall accuracy might deny 60% of qualified applicants from underrepresented groups, and a medical diagnostic model with 0.95 AUC might have near-random performance on…
Premium Prompt
Unlock this prompt — and all 30,000+ expert-crafted prompts — with Pro.
Unlock with Pro