Prediction of Loan Approval in Banks Using Machine Learning Algorithm

Normally, the banking sector gains profit from different kinds of sources, but the main factor in their profitable growth is giving loans to their customers. A bank's profit or loss is determined by the quantity of the loan. A huge number of people are taking loans from banks and not clearing their dues. So there is a tedious process for banks to identify them and collect them. This in turn leads to following up on a complex verification process and issuing loans to their customers. Customer satisfaction is mandatory in every field, but in the banking sector, it is essential to identify genuine applicants and satisfy their needs, as well as to keep bankers free from non-payable customers. The manual method of identifying the right person and providing a loan to them may cause misconceptions. Several studies in this field suggest various kinds of approaches to identifying applicant status. Therefore, the use of machine learning (ML) approaches in the banking sector to resolve difficulties faced by bankers is becoming more popular. This reduces the problems faced by bankers and makes their applicants satisfactory. The time taken for this process is too long. It is essential to select the best ML approach from the different kinds of classifier algorithms available. The attributes chosen for this study include the applicant's education, type of work, applicant income, loan amount, co-applicant income, credit history, and property area. The proposed study reviews ML approaches such as K-nearest Neighboring (KNN), Logistic Regression, Decision Tree and Support vector Mechanism (SVM). Among those, KNN outperforms the other three algorithms, and the efficiency obtained by KNN is 87.50%, Logistic Regression is 83.512%, Decision Tree is 77.09%, and Random forest (RF) is 74.49%.