Development Prediction Model to Optimize Cooperative Loans Based on Machine Learning Algorithms

Abstract

Default on loans by borrowers to the cooperative to optimize the cooperative's business performance. In this research, a default prediction model was developed using several quite popular machine learning algorithms, namely decision tree, K-NN, logistic regression, and random forest, then all models with each of these algorithms were compared and evaluated. to find out which algorithm model is the most effective and accurate in predicting loan defaults in cooperatives. Model evaluation is carried out using metrics such as accuracy, precision, recall, and f1-score. The dataset used in this research was obtained from the loan list at one of the Savings and Loans Cooperatives in Tasikmalaya Regency, the contents of which include attributes such as borrower profile, loan amount, number of installments, and others. This dataset is divided into training data and test data to train and evaluate the model. These machine learning algorithms were chosen because they are quite well known among other algorithms for prediction and have been proven in several financial studies. The results of this prediction model can be used by cooperatives to support decisions in providing appropriate loans.