| Summary: | 碩士 === 銘傳大學 === 管理科學研究所 === 91 === Taiwan banking industry faces the trend of executing of Financial Control System with merging banks and diversity financial products. Under the serious competition environment, the bank must create the competitive advantage. Therefore, that the bank increase the customer’s loyalty to sustain the customers by Customer Relationship Management（CRM）. From two major banking businesses of saving accounts and credit accounts, the most important thing for the bank is to delineate current valuable customers. Effectively segment current customers will help the bank to sustain the valuable loyal customers to increase the profits by the CRM.
We classify holders of the credit cards by some variables such as 「holding time」,「using rate」,「average money」evaluating the degrees of the behavioral loyalty with four types：the most loyal, loyal, potential and silent holders.Classifying the account of the account bank database，We use behavior loyalty and valuable customer to classify with four types：the best, valuable, loyal and lost account. Establishing the two classifying models with discrimination analysis, back-propagation neuron network and decision tree, we compare three means and find that back-propagation neuron network and decision tree are better in this paper.
Since joining the WTO, Taiwan winery industry also face much competitive pressure of entering foreign winery. Therefore, how to keep to the market share rate, diversify the winery products, and create the profits are the most important issues for Taiwan winery industry.
As for the retailing industry mining, the study uses the selling data of one wine agent as the case study to decide the optimal selling package and the possible recommended cross selling merchandises by the basket analysis of customers’ purchasing behavior and preference of use the basket analysis. The mining result revealed that product B700ML can sell together with all kinds of products as the best selling mix. Because of the different distribution characteristics of cross selling and associate rules of distribution channel, most distribution and best associate rule include product mix of P750ML and S750ML. So, P750ML and S750ML are viewed as the best cross selling product mix.
|
|---|
