| Summary: | 碩士 === 國立高雄應用科技大學 === 商務經營研究所 === 96 === Following the credit card and cash card crisis in Taiwan in 2005, all banks has pulled out from the cash card market, except for China Trust and Cosmos Banks. Despite the lucrative rewards offered by the cash card market which prompted the re-entry of the two banks, the risk management will remain the single most crucial subject. Subsequent to the tighten control from the authority following the crisis, issuance banks no longer targeting on the volumes of the cards issued. Instead, the real gain comes from effective risk management and focusing on customers with great potential.
This research aims to explore and discuss how the banks may use the existing customers information database to establish a credit scoring model in order to shortlist good graded customers. Based on the customers database from a local bank, we adopted Logistic Regression and Decision Tree methods to form a Credit Scoring Model. Comparisons are then made to select the most suitable method. This experiment showed that both the methods have better projections when there are external variables available as compared to just internal variables available. However, Decision Tree method yielded a more accurate projection when compared to Logistic Regression method. Hence we have selected Decision Tree method as the Credit Scoring Model to be used, for it being easier to understand and to apply.
As cash card is a high risk and high rewards product, our research provide the basis of selecting potential customers for further marketing, which reduces unnecessary costs and therefore increase the profit margin. An added value to this exercise is that through the analysis, the bank can further sub-categorise its customers base into more specific groups and therefore further analysis and marketing are possible. This enables the bank to manage its customers effectively, and to maximize its earnings.
|
|---|
