Predictive Creditworthiness Modeling in Energy-Saving Finance: Machine Learning Logit and Neural Network

Customer's creditworthiness was becoming more crucial for ESCO. Machine learning used to predict the of clients in retrofit financing processes. ESCO This research aimed develop a retrofitting scoring model leverage machine and life cycle cost analysis (LCCA) evaluate alternative Energy Efficiency Saving Indonesia. The built on Logistic Regression Artificial Neural Networks learning. developed tested using Python algorithm, proposed model's efficiency demonstrated. logistic regression calculations showed that accuracy value prediction data with test 88.3562 % 87.67% models. rate result refers correct predictions among all 92.20% 91.98%, respectively. Meanwhile, percentage customers who were predicted default 94.41% 93.81% model. Credit models helpful risk consumer loans. Finally, quality performance these evaluated compared identify best one. neural network obtained good very similar, although slightly better.