A Random Forest Classification Model to improve credit evaluation in Credit Unions - Perú

Coquis A.; Sonan C.; Rodriguez D.; Vargas A. · Proceedings - 2025 3rd International Conference on Intelligent Control and Computing, IC and C 2025 (2025) · DOI: 10.1109/ic-c68228.2025.00030

Resumen

Credit Unions are non-profit financial institutions that operate in a comparable way to a bank. These organizations currently have problems being able to accurately assess the creditworthiness of their members. Therefore, it is proposed to develop a Random Forest Classification model that qualifies the eligibility and payment capacity of the member to receive a loan using the information collected in their application and thus increase the quality loans that the Credit Unit deliver while freeing up the burden on the credit unit. To choose the Machine Learning model, an evaluation of the most used algorithms in similar applications was carried out. The model was trained and tested with the data of the financial institution. The implementation and its correct operation were validated through an experiment in which fifty evaluations were carried out through expert judgment and compared with the same cases evaluated through Random Forest, reaching an accuracy of 95.56% and saving them twenty hours of work for every fifty evaluations.

Instituciones: Universidad Peruana de Ciencias Aplicadas