چکیده:
This paper develops a decision support tool using an A2 method and data envelopment analysis (DEA) approach (A2-DEA). This new method is applied for the bank credit customer selection problem and credit scoring as a pilot survey at Export Development Bank of Iran. The proposed method has led to fewer calculations, faster and more accurate decision making, less complexity, and ability to analyze many scenarios with only one or a few judgments of decision makers while the effect of the subjective opinion of one single decision maker will be avoided. This proposed method is compared with adaptive analytical hierarchy process approach, which is suggested by Lin et al., in 2008, and it is named A3. An illustrative example demonstrates the implementation of the proposed approach. This example demonstrates how this approach can avoid the main drawback of the current method, and more importantly, can deal with the credit customer selection more convincingly and persuasively. The implementation results show that this method is significantly valid for ranking credit customers. Comparison of methods shows that although A3 have benefits, it also suffers from limitations, which can be avoided by the A2-DEA model, also improves the time and cost needed for implementing in comparison.
خلاصه ماشینی:
"This approach uses the result of AHP and DEA as input values of the Artificial Neural Network (ANN) model for developing and improving the A2 method, with a little change based on the experience of experts in the field.
In the proposed model, when AHP- DEA and ANN models are set up, measuring the priority of industries for issuing loan when the budget is limited, can be made easy by the opinions of a few decision makers, thus the calculation of the geometric mean of answers that are obtained from many experts will be unnecessary.
Experimental details of proposed model (A2–DEA model) Step 1: Collecting data and calculating the weights of the criteria Collecting the expert judgment for running AHP-DEA model is very important and because of the large number of decision makers, it is a challenging work.
Step 3: Ranking of credit customer by ANN Adopting a global approach in presenting the data to the network, the ANN model for the measuring the rank of credit customer in different industries sections was trained by BP algorithm based on Levenberg - Marquart rule.
In this study, using the AHP model proposed by Herrera-Viedma, it is possible to avoid checking consistency and collecting the experts' opinions in less time, so that the number of pairwise comparison judgments is declined to n-1 whereas the traditional analytic hierarchy approach uses n(n-1)/2 judgments in a preference matrix with n attributes."