Abstract
Credit scoring is a decision-making task concerned with how to assign credit applicants to one of two groups: a "good credit" group that is likely to repay the financial obligation, or a "bad credit" group that should be denied credit due to his/her high likelihood of defaulting on the financial obligation. Inappropriate credit scoring models may increase the cost of credit analysis and delay credit decisions. Moreover, such models may be ineffective in monitoring existing credit accounts. The objective of this research was to investigate the methodologies to mine customer credit history for the bank industry. Particularly, combination of logistic regression model and neural network technique are proposed to determine if the predictive capability of neural network models or logistic regression used in credit scoring can be improved. To demonstrate the effectiveness of the proposed combined approach, the technique is applied to the credit data from one large bank in Taiwan. The backpropagation learning technique with various learning rate is extensively studied to determine the connection weights between neurons. In addition, the number of hidden neurons is also varied to see its effect on the converge rate. The extensive studies were performed on the robustness of the built network model in terms of different training and testing sample sizes.
Kao, Ling-Jing (2001). Mining customer credit by using neural network model with logistic regression approach. Master's thesis, Texas A&M University. Available electronically from
https : / /hdl .handle .net /1969 .1 /ETD -TAMU -2001 -THESIS -K365.