Due to the wide availability of computers, information, and communication technologies data are being generated massively today, especially in financial institutions and banks data are being generated massively on a regular basis. Microfinances are one such institution that collect, process, and store huge amounts of records from time to time and therefore deal with large amounts of data. On the other hand, Microfinance is facing problems in loan risk assessment and managing portfolios at risk. Currently, microfinance institutions’ loan risk assessment and granting loans to the borrowers is conducted traditionally depending on the loan approval team’s views and beliefs, Moreover, such a way of risk assessment creates inefficiency in the quality of identifying borrower’s characteristics before granting the loan. If microfinance institutions (MFIs) do not manage their loan risks well, they are likely to fail to meet their social and financial objectives. The existing past and historic data related to loan borrowers and loan characteristics could be actionable and usable for loan risk assessment with the help of Machin learning algorithms. This study was conducted to demonstrate the practical methods, experiments, and datasets with machine learning to assist MFIs through building a classification and prediction model which supports the prediction of a new loan borrower’s status (Active or Defaulter) when the loan decision-making in the microfinance institutions. The classification and prediction model is built based on the MFIs loan borrowers’ data. Necessary preprocessing activities have been applied to clean and make it ready for Experimentation. Then, the four algorithms used were SVM, KNN, Naïve Bayes, and logistic regression. The RStudio with R programming was used to simulate all the experiments. A confusion matrix was used to calculate the accuracy, specificity, sensitivity, and precision were used to evaluate the performance of the models, and a Cross table was used to visualize the performance of the models. The results of the experiment show high precision so that the models can be used in detecting and predicting defaulter (risky) loan applicants. The KNN classifier produced an accuracy of 99.91%, the SVM classifier produced an accuracy of 92.4%, the logistic regression model also produced an accuracy of 93.8%, and the Naïve Bayes classifier produced an accuracy of 83.8 %.
The Loan Risk Prediction Using Machine Learning Algorithms: -The Case of Micro-Finance Institution’s, GET MORE COMPUTER SCIENCE PROJECT TOPICS AND MATERIALS