hybrid method for customer credit scoring

Appropriate customer selection is a key element of risk management in the banking industry. however, achieving accuracy in risk assessment is considered a difficult issue. predictive analysis for credit risk primarily calculates a loan applicant s likelihood of default based on their demographic and personal data. manual procedures are finding it difficult to handle the enormous volume, partiality, and variety of credit data. therefore, in this study a hybrid method is used in practical applications to handle this challenge. the proposed model is a deep neural network architecture composed of convolutional neural networks (cnns), long short-term memory (lstm), and fully connected layers (dnn), culminating in a softmax classifier for binary classification tasks. we used the “default of credit cards clients dataset” collected by uci machine learning repository. the accuracy of proposed method is 86% which is higher than previous researches.