碩士 / 輔仁大學 / 統計資訊學系應用統計碩士在職專班 / 104 / Under fierce competition among retail industry market, each retailer continuously introduce varies promotions to attract customers in order to increase sells. By using coupons, in particularly is one of the most common promotion method. However, excessive promotion may lead to consumers’ paralysis, and become ineffective promotion. Therefore, how do retail industries use scientific methods to identify high probability of repurchase customers, and issue them relevant coupons are most important topics of this study.
In order to discover a customer with a higher rate of repurchase effectively and scientifically, the study utilize data mining techniques from massive transaction data, jointly identify consumer behaviors, and derived highly correlated continuous and categorical variables as model variables of repurchase. Afterward, use logistic regression to build two-types-variable prediction model separately. After comparison of those prediction models, the study has confirmed that through predictive model has more efficient and effectively to identify those repurchase than using translation random methods.
The prediction model has correctly predicted rate of 61.71% from the study. Up to 69.85% of target customers, and 56.7% of non-target customers. The results illustrate the industry will increase 1.5 times more effective to find the target audience. The prediction mode can benefit the industry to understand the characteristics of potential repurchase, it also can assist the industry to filter out possible repurchase customers for precision marketing, marketing strategy, reduce cost and improve the relationship with customers, to end with create higher profits.
| Identifer | oai:union.ndltd.org:TW/104FJU01506014 |
| Date | January 2016 |
| Creators | LIN,TZU-YAO, 林資堯 |
| Contributors | LIANG,TE-HSIN, 梁德馨 |
| Source Sets | National Digital Library of Theses and Dissertations in Taiwan |
| Language | zh-TW |
| Detected Language | English |
| Type | 學位論文 ; thesis |
| Format | 100 |
Page generated in 0.0015 seconds