A Data Mining Based Approach To Customer Response Model — A Case Study

• Main Authors: Shu-Chen Kao, 高淑珍
• Other Authors: Hae-Ching Chang
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/24009259769804126741

博士 === 國立成功大學 === 企業管理學系碩博士班 === 92 === It has been seen that the modern marketing paradigm has been rapidly shifting and business has used to apply target marketing to capture the right customers in promotion activity. However, the customer response model, regarded as the tool for targeting and prediction, is the most important task in marketing promotion. This research proposes a data mining (DM) based customer response model for insurance industry to help in finding unobvious but valuable promotion knowledge to support making marketing related decisions. First, we visited a leading insurance company (denoted by A), one of the most popular insurance companies in Taiwan, to frame the research focus. There were 188464 transaction records provided by A company. Of the collected data, two to third was used as a dataset being mined while the remainder as a test dataset. The ID3 mining algorithm was utilized to derive decision rules and obtained 943 qualified rules in total. The accuracy of the proposed model was 81%. To capture the important implication of the knowledge, the research analyzed the obtained rules in two directions including the level of supports and degree of conditions. The former focused mainly on the amount of supports and degree of conditions for the obtained rules to analyze the product categories with respect to the customer characteristics. The latter carried primarily out the relationships between different degree of condition and product categories. The research then conducted the second visit of A in order to validate the obtained knowledge in practice. The results indicated that the customer response model was able to aim at finding and diffusing the insurance marketing knowledge. It was also found that the proposed customer response model with the DM mechanism was decision-supportable based on the opinion of executive manager. Moreover, the proposed model would play a key role in changing the decision making style from experience-oriented to information-oriented. Other research findings were provided and managerial implications addressed in this research also.