PEC Framework: Cross-Domain Validation Datasets (Climate Risk, Fintech Credit, and Health Outcomes)

 This dataset collection supports the cross-domain validation of the Predict-Explain-Certify (PEC) framework on three external domains beyond the original Madagascar food security application. Three datasets are       
 included:                                                      
                                                                             
   (1) Climate Risk — Southeast Asia: 120 observations, 30 features, 12 regions (Vietnam, Cambodia, Laos), 10 years (2014–2024). Targets: Flood Risk Index (FRI), Economic Damages (ED), Population at Risk (PAR).          
 Certification: 89.6 (B), 96.5 (A), 99.2 (A).                                
                                                                             
   (2) Fintech Credit Risk — West Africa: 135 observations, 27 features, 15 countries (WAEMU), 9 years (2016–2024). Targets: Default Rate (DR), NPL Ratio (NPL), Fintech Penetration (FP). Certification: 87.4 (B), 88.9 (B),   
 98.9 (A).                                                                   
                                                                             
   (3) Health Outcomes — East Africa: 140 observations, 31 features, 14 DHS regions (Kenya, Tanzania, Uganda), 11 years (2013–2023). Targets: Infant Mortality Rate (IMR), Vaccination Coverage (VC), Qualified Deliveries (QD). Certification: 97.7 (A), 98.9 (A), 99.2 (A).                                

Ridge regression wins 9/12 targets, confirming the interpretable model advantage. Multi-method triangulation (SHAP + LIME + Permutation Importance) achieves Kendall τ > 0.75 for all domain-target pairs. Temporal 
 features dominate SHAP in all 9 targets. Three domain-specific adaptation requirements are identified: fairness weight rebalancing for financial  services (w_T=0.4), relaxed stability thresholds for climate data (CV<15%), and AUC metrics for imbalanced classification. Operational platform:  https://pec-gen.streamlit.app