在本研究中,為了幫助行銷人員選擇eDM發送的目標客戶,我們利用概念漂移建構的會員回購力指數,以預測實體超市會員的回購力並選取高回購力的目標客戶。根據會員預測的高回購力,行銷人員可以為不同類型的超市會員設計適當的促銷方案,以提高促銷效率。 由於我們的研究的對象是台灣某知名超市,而超市主要的販售商品以食品居多,食品屬於消耗品,導致顧客下次消費時仍然會重複購買相同的商品。另外,由於顧客在選擇自己喜歡的產品之前只會花些許的時間來閱讀eDM,造成eDM的商品數量只能包含一小部分的固定商品。因此在本研究中,我們將考量以上的特性提出一個新的商品組合推薦計算方式,針對不同會員設計不同商品組合的eDM。同時,在本研究中,我們使用這家超市的顧客交易記錄作為測試數據。這家超市的行銷部門審查了所有的實驗結果,並證實我們的方法不僅超越了超市採用的現有方法,而且還有助於超市顧客的選定及設計合適的eDM促銷策略。 未來,可依概念漂移選取不同回購力的顧客,並且亦將產品結合價格,讓行銷人員設計最佳的促銷策略,以提高超市營收。並可結合分析顧客線上消費行為、線上瀏覽行為與商品購買週期,對顧客進行線上線下資料整合分析,以祈未來能在適當的時間,選取適當的人選,推薦適當的商品,將能大大提升顧客來店消費之意願,吸引顧客回到門市進行消費。
In this study, To help marketers choose the target customers for eDM, we target customers with high repurchase power. Therefore, We use the customer repurchase index with concept drift to predict members’ repurchase rates for a physical supermarket’s members and select target customers with high repurchase power. Based on members’ predicted repurchase rates, marketers can design proper marketing strategies for different types of supermarket member to improve marketing effectiveness. Becasuse our research target is a well-known Taiwanese supermarket, suprtmarket mainly sell food, and the food is consumable. Customers will still purchase the same commodity repeatedly when next time back to store. In addition, because customers can only spend a short amount of time reading eDM before choosing the commodity that they like, In this study, a new recommendation system for the most suitable combination of cmmodity under the above condition that a customized eDM can include only a small, fixed number of such types. In this study, customer transaction records from a well-known Taiwanese supermarket were used as the test data. This supermarket’s marketing department reviewed all the experimental results and confirmed that our approach is not only superior to the current approach employed by the supermarket but also useful in designing appropriate eDM marketing strategies for selected supermarket customers. In the future, customers with different repurchase powers can be selected according to the concept drift, and the commodities are combined with the price, so that the marketing staff can design the best promotion strategy to increase revenue. It can also analyze the online consumer behavior, online browsing behavior and commodity purchase cycle to recommend appropriate commodities at the right time to selected target customers for attracting target customers back to the store for consumption.