MACHINE LEARNING ALGORITHM IN CREDIT SCORING TO PREVENT BAD DEBT IN COOPERATIVES
This research proposes a credit score model for cooperatives using machine learning. Until now, there is no standard credit score assessment in savings and loan cooperatives in Indonesia. There are still many savings and loan cooperatives that provide loans due to closeness to the management or manager of the cooperative. The purpose of this research is to obtain a credit scoring method through machine learning that is effective, efficient and high accuracy. To predict the chance of default, this research uses seven machine learning algorithms namely Logistic Regression Classifier, Support Vector Machine Classifier, K-Neighbors Classifier, Decision Tree Classifier, Random Forest Classifier, XGBoost Classifier, and Light Gradient Boosting Machine Classifier. The data taken from the loan data of 851 members of Bank BPD Jateng "Yakekar" Cooperative, Semarang, Indonesia. The results show that Logistic Regression, Support Vector Machine Classifier, and K-Neighbors Classifier are the models that have relatively better performance in identifying 'Current' collectibility. However, all models have difficulty in classifying other collectibility ('Bad' and 'Doubtful') with low precision and recall.
Machine Learning; Credit Scoring; Cooperatives.