The Research of Applying Improved Artificial Neural Network to Credit Card Customer Relationship Management

• Main Authors: Yu, Huey-Hwa, 俞慧華
• Other Authors: Wu, Chung-Min
• Format: Others
• Language: zh-TW
• Published: 2002
• Online Access: http://ndltd.ncl.edu.tw/handle/95161547377972627335

碩士 === 國立臺北科技大學 === 商業自動化與管理研究所 === 90 === In Taiwan, the using amount of credit card and outstanding credit has increased rapidly and the competition of banks turns white-hot; Revolving interest, mail-order and so on, bring the major benefit for credit card business. For these reasons, detecting probability of bad debt actively, identifying customers of higher profit return, and increasing customer’s loyalty, is the best direction of bank strategy. This research proposes a method to select variables for classifying models based on Genetic Algorithms, to improve the efficiency of Artificial Neural Network. Conclusions obtained credit line, number of credit cards, 7th of industry, 1st, 3rd , 4th of work area, 1st, 2nd, 4th, 6th, 7th, 8th of professional title, level of education, are the most important factors of forecasting quality. These factors totaling thirteen variables are 37.1% of original variables. In our research, we decreased variable numbers from 35 to 13, and decreased time for trying errors and collecting data substantially, but we still had 91.17% correct rate.