Five-Category Evaluation of Commercial Bank's Loan by the Integration of Rough Sets and Neural Network

This article proposes a hybrid intelligent system that predicts five-category risk grades by feeding past financial performance data into rough set approach and neural network. The attributes are reduced with no information loss through rough set approach, and then this reduced information is used to develop classification rules and train Elman neural network. For the experiment, the financial data of 896 firms are selected to predict risk grades. The effectiveness of our methodology is experimentally verified by comparing traditional logistic model with our hybrid approach.