Decision Support for Feasibility Evaluation of Ecology-Oriented Development Projects: Integrating Case-Based Reasoning and Machine Learning Approaches

Ecology-oriented development (EOD) projects integrate ecological governance with revenue-generating industries, but their financial feasibility evaluation and resource balancing depend heavily on expert judgment. This study develops a decision-support method combining case-based reasoning (CBR) and machine learning (ML) to support the systematic reuse of historical project experience. A standardized case representation framework and an EOD case database were established. Gradient boosting decision trees (GBDT) were then used to determine predictive-importance weights, support weighted similarity calculation, and predict key financial indicators, while CBR linked retrieved case information to project-specific recommendations through experience transfer and adaptation. The method was applied to the Xibei Canal EOD project in Wuxi, China. The three most similar historical cases achieved similarities of 91.02%, 90.26%, and 89.52%, respectively, providing complementary references for financing arrangements and industrial cross-subsidization of the Xibei Canal EOD project. GBDT achieved R² values of approximately 0.87, 0.97, 0.93, and 0.97 for predicting financial internal rate of return (FIRR), financial net present value (FNPV), payback time (Pt), and debt service coverage ratio (DSCR), respectively, providing quantitative support for scheme evaluation. Threshold sensitivity analysis further clarified the scope of reuse-eligible cases and the corresponding adaptation requirements under different screening conditions. The proposed method provides structured, quantitative, and traceable support for EOD project screening, preliminary financial assessment, resource balancing, and scheme comparison.