Bank Repayment Prediction-System On Deep Learning Techniques

The loan is one of the essential products of financial institutions.All institutions are trying to develop effective business strategies to encourage more customers to apply for their loans.Many people are applying for bank loans,but the bank has limited assets that need to be granted to a limited number of people only,so finding a safer option that can be granted to the bank is a typical process.Some customers are not in a position to pay off the loan after their application has been approved.Consequently,financial institutions are faced with a decision where the default application increases.Also,the traditional machine learning techniques are not very effective and efficient,such as supporting vector machine(SVM),which requires a large amount of memory and computing time to handle huge data sets.Using the well-performing KNN approach to solve this problem can suffer less accuracy than SVM because of its low efficiency while executing the methods and dependency on choosing a "good value" fork.The original random forest algorithm has some limitations in selecting features due to selecting significant classifier numbers,the number of random training features,and the combination steps.This research aims to propose an effective,accurate,and scalable predictive bank repayment system based on deep learning that addresses all of the above-listed issues.The key contributions to developing the proposed method,namely using the Deep Neural Network to predict bank repayment status,are that a more robust process has been built and can effectively manage various kinds of data issues in the banking systems.An excellent alternative to using efficient deep learning techniques rather than traditional approaches to machine learning.Therefore,we try to reduce this risk factor by selecting the right person and saving lots of the banks' effort and assets.Using our proposed system for bank repayment status,banks can reduce the number of bad loans and losses.The proposed method addresses different types of problems in the banking sector,e.g.,the system cannot process huge data.The learning techniques applied in the system are not very useful,and different data parameters cannot be managed.KNN algorithm accuracy depends on the quality of the data,and with large data,the prediction stage might be slow.Sensitive to the scale of the data and irrelevant features.KNN required high memory need to store all of the training data,and given that it keeps all of the training,it can be computationally expensive.Therefore,the deep learning methods achieve the desired outcome for our proposed prediction of repayment of banks.Deep learning's primary advantages over other machine learning algorithms are its capacity to execute feature engineering on its own.A deep learning algorithm will scan the data to search f or features that correlate and combine them to enable faster learning without explicitly telling them.The reason behind the boost in performance from a deeper network is that a more complex,non-linear function can be learned.Given sufficient training data,this enables the networks to more easily discriminate between different clas-es.The proposed model delivers superior results in this research using advanced deep learning approaches,and also helps banking institutions reduce potential losses and make better predictions for predicting a business's new loan applicants.Using the proposed model,it gets the features automatically deduced and optimally modified for the desired outcome and automatically learned various types of variations in the data.As we compared the findings with traditional models,the accuracy of Deep Neural Network is 82%,which is improved by 13%from Random Forest,which is the highest accuracy from the conventional model.Random forest accuracy is 69%.In this study,we have used loan application data with one lac(100,000)records taken from Kaggle,and the experimental results show satisfactory performance using the proposed bank repayment prediction system.Eventually,this new model will help predict the future consumer loan book and its repayment status to find good and bad customers.