The most critical issue for financial and entrepreneurship institutions is determining customers' ability to repay the facilities. Furthermore, the customers' success in running and fulfilling their business should be considered. A wide range of research studies has been carried out on customers’ authentication about granting personal credit cards and consumer loans to them. The issue that has remained untouched or less investigated is that most loans paid for starting up a business encounter the problem of not setting up that business; moreover, applicants may fail to maintain their business after launch. These issues can result in several issues such as inflation in society, non-growth of assets and stalled transactions, and huge losses for loan lending institutions. One appropriate solution to overcome such problems is to select proper applicants for granting facilities with minimum failure probability for project fulfillment or facility non-repayment. From this standpoint, the present study presents a decision-making model for the automatic selection of facility applicants which is a critical need in financial institutions. For this purpose, a relatively comprehensive and complete dataset based on existing standards is generated. This dataset is designed by considering the patterns of prosperous and non-prosperous customers in four areas of prediction: supervision status, facility status, number of generated jobs, and activity time duration. A deep feedforward neural network (DFNN) model is designed and developed to extract features from input data. The proposed DFNN model is tested and investigated not only on the generated dataset, but also on other datasets, and it is then compared to other methods available. It is observed that the proposed model is significantly successful and efficient.
展开▼
