An Explainable and Multidimensional Climate Performance Index: Integrating Statistical Validation and Machine Learning-Based Structural Diagnostics

Assessing climate performance through emission-centric metrics provides an incomplete picture of countries’ progress toward integrated climate objectives. This study proposes the Climate Integrated Performance Index (CIPI), a multidimensional and explainable composite indicator for 27 European countries during 2015–2023. CIPI integrates six thematic dimensions: emissions, energy systems, mitigation capacity, transport, agriculture, and waste–land-use interactions, using robust normalization, a policy-informed weighting framework, and formal statistical validation. Based on 243 country–year observations, the results indicate that CIPI is non-redundant. Pearson correlations reveal strong positive associations with the Energy Index (r = 0.899) and Mitigation Index (r = 0.894), alongside a significant negative association with the Agriculture Index (r = −0.659), highlighting sectoral trade-offs. Variance decomposition further shows that energy and mitigation dimensions jointly account for approximately 87% of explained variance, whereas agriculture exerts a systematic counterbalancing influence. To support structural interpretation, an explainable machine learning framework combining XGBoost and SHAP was implemented as a diagnostic layer. Renewable-energy capacity emerged as the dominant structural driver of integrated climate performance, and SHAP-based analyses revealed a nonlinear threshold effect, with positive contributions accelerating beyond a normalized renewable-capacity level of approximately 0.58 (95% bootstrap confidence interval: 0.54–0.62), particularly under low fossil-fuel dependency conditions. Because the machine learning models use indicators that also contribute to index construction, the results are interpreted as evidence of structural consistency and diagnostic interpretability rather than independent predictive discovery. To address this limitation, repeated cross-validation, subsample validation, benchmark comparisons, and weighting-sensitivity analyses were conducted. Ranking robustness remained high under alternative weighting schemes (Spearman ρ > 0.96), while comparison with an emission-centric benchmark demonstrated substantial rank reversals, indicating that broader sectoral and policy dimensions influence climate-performance assessment. Overall, CIPI functions not only as a benchmarking tool but also as a transparent diagnostic framework for identifying structural trade-offs, nonlinear relationships, and policy-relevant climate-transition dynamics.