A PCA-based composite sustainability measure for maritime logistics: A diagnostic machine learning assessment for Türkiye
Gencay Sariisik, Yasin Ozkan, Betul Goncu
article (Uluslararası, Hakemli) | Yazar(lar): Gencay Sariisik, Yasin Ozkan, Betul Goncu | Yıl: 2026 | Dergi/Kurum: RESEARCH IN TRANSPORTATION BUSINESS AND MANAGEMENT | İndeks: WOS.SSCI | DOI: 10.1016/j.rtbm.2026.101825 | Anahtar Kelimeler: Maritime logistics, Sustainability measurement, Principal component analysis, En
This study develops a composite sustainability measure for T & uuml;rkiye's maritime logistics sector by integrating Principal Component Analysis (PCA) with diagnostic machine learning techniques. Using a framework built on standardized environmental, energy, and operational indicators for the period 2013-2024, the study constructs a relative sustainability benchmark designed for within-country performance assessment. The first two principal components explain 74.64% of the total variance, indicating that carbon emissions and energy-use patterns constitute the dominant structural dimensions of sectoral sustainability performance. Ensemble learning models, including Gradient Boosting (GB) and Extreme Gradient Boosting (XGBoost), are employed as nonlinear diagnostic tools, not to generate forecasts but to evaluate the internal consistency and sensitivity structure of the PCAderived index. The high explanatory performance observed across these models primarily reflects the structural coherence of the index rather than independent predictive capability. SHAP (SHapley Additive exPlanations)based interpretation shows that emission-intensive variables exert the strongest marginal influence on sustain-ability scores, while improvements in energy efficiency and selected operational indicators contribute positively. Taken together, the results identify decarbonization and energy optimization as the most influential drivers of improved sustainability performance in T & uuml;rkiye's maritime logistics sector. The proposed framework offers a transparent and replicable measurement approach that can support sustainability monitoring, Environmental, Social, and Governance (ESG)-oriented decision making, and evidence-based transportation policy evaluation in maritime logistics.