Investigations on Classification Methods for Loan Application Based on Machine Learning

Abstract

As there is an increasing trend of people consuming by debit in China, financial organizations deal with a lot of loan applications. If customers cannot repay the loans on time, the organizations have to cover the loss. Therefore it is important to predict correctly whether a customer will repay the loan on time. Typical machine learning methods can be employed to exploit customers' financial information and give valuable judgements. We investigated the function of Deep Neural Network (DNN) in this work, as it achieves high successful rate in fields of image recognition, speech recognition and natural language processing. We compared it with traditional learning methods, such as Naïve Bayes, decision tree and K-Nearest Neighbor. Experiments showed that DNN achieves better performance than its traditional competitors. The accuracy and recall of DNN are 0.73 and 0.42 respectively. Its It-score is 25% higher than the best one of traditional methods.