Abstract:
Saving and credit cooperative plays an important role by delivering loans and saving services to the peoples especially low and middle income people which are not accessed by the bank. Loan repayment is the big challenge in saving and credit cooperative institutions. The objective of this study is to develop a model to predict loan repayment performance of the customers. In order to perform this research, Hybrid DM process model was used to explore data mining processes to find interesting patterns by using python programing language.
To perform this study, the datasets were taken from 4 years saving and credit cooperative datasets with a total of 11928 instances and the most appropriate attributes were selected using feature selection techniques for experimentation to build the model such as level of education ,number of penalty, purpose of credit, job, loan repayment period ,guarantee, duration of membership, age, saving amount, loan amount) from those attributes the researcher used new attributes like (job, level of education, Guarantee, and number of penalty. To achieve the objective of these research different experiments were conducted using MLPANN, Ensemble and Decision tree classifiers algorithms. In addition, the predictive performances of the classifiers were evaluated and compared using accuracy rate, as TP rate, precision, recall, confusion matrix and ROC curve. Based on evaluation metrics, out of the three classifiers Decision tree classifier has best accuracy and it achieve 98.3 %. As a result Decision tree classifier was selected for implementing the model to predict the loan repayment performance of customers.
Key words: loan repayment status, saving and credit cooperative, MLPANN, Ensemble and Decision tree algorithm
