Performance Analysis of Loan Classification for Commercial Banks in Neural Network

Abstract

With the rapid growth in banking services, there has been a tremendous increase in the number of individuals and businesses applying for loans. It is therefore getting tougher and tougher for banks to make correct and consistent decisions regarding loan approval. In this regard, Neural Networks (NN) can play an important role in financial institutions for such tasks of loan classification and making decisions about loan sanctions. This study develops more accurate Multilayer perceptron (MLP) as an enabling tool to support loan decisions in commercial banks analyzing different features of loan applicant. The dataset consists of different representative cases of loan applications that were considered or rejected based on the guidelines of banks, to train and validate the neural network model. The proposed study shows the effectiveness of the neural networks under balanced datasets which can play an important role to understand the impact of quality of dataset as well. It illustrates the ability of neural network model to predict the creditworthiness of an application accurately and precisely preventing the bank and its officials from making erroneous decisions with regard to loan approvals. The proposal aims to shed light on the exploration of the available datasets, selection of the appropriate neural network and using them for making correct and consistent loan decisions. The main goal is to create an accurate deep neural network that will take into consideration all independent variables and based on that will predict if the applicant is going to get loan approval or not. Also working of the proposed model will also be compared with other classifiers such as KNN (K Nearest Neighbor) Classifier and SVM (Support Vector Machine) classifier in terms of accuracy, sensitivity and F₁-scores.