網路科技日益發達，透過電腦、手機或其他電子設備可以輕易地連上網路，網路購物也順勢成為了大多數人們生活中不可或缺的交易方式。過去已有相關研究僅須顧客歷史之交易日期及交易金額，透過計算出一組可描述顧客總體交易行為的參數(a,b,r,s)，即可計算每一顧客未來某一期間之活性機率。亦有學者使用Bootstrap隨機抽樣方法預測購客未來購買行為。但預測準確率並不穩定，原因可歸於參數並非最佳之參數，及未考慮顧客最後購買日，導致管理者在決策上的困擾。基於此，本研究使用一個簡單的方法－RFM，企業同樣只需要提供顧客編號、交易日期及交易金額，即可計算足以充分描述該名顧客購買行為之欄位，並以隨機森林模型計算出穩定且精準的結果，供管理者決策上使用。我們使用台灣知名購物平台的實際資料模擬預測結果，預測結果與先前研究比較之後，各項指標均有相當大的突破。

Internet technology is rapidly developing. We can easily connect to the Internet through computers, mobile phones or other electronic devices. Online shopping has also become an indispensable trading method in most people’s lives. There have been relevant studies in the past which only require the customer’s historical transaction date and transaction amount. By calculating a set of parameters (a,b,r,s) that can describe customers’ transaction behavior, each customer's activity rate in a future period can be calculated. Some scholars use Bootstrap random sampling method to predict the future transaction behavior of each customer. However, the results were not stable. The reason can be attributed to the fact that the parameters were not the most suitable, and the last transaction date of each customer was not taken into consideration, which leads to the confusion of managers in decision-making. This study uses a simple method – RFM. After comparing the predictive results with previous research results, each of the indicators has made a considerable breakthrough.

中文部分：
布萊恩‧克里夫頓、《透視數據下的商機：運用Google Analytics發掘商業洞見》、天下雜誌、台北、2016年。
何佩珊、《沒它不行，東森集團全通路布局背後的關鍵大腦》、數位時代、台北、2016年。
陳傑豪、《大數據玩行銷-從舊4P到新4P，從大數據預測下次購買時間》、30雜誌、台北、頁59-69、2005年。
陳昇瑋、《讓資料為你產生價值》、哈佛商業評論、129期、頁120-125、2017年。
葉迪珺、《以Bootstrap來建立一個穩定的忠誠客戶預測模型》、國立成功大學會計系研究所碩士論文、2017年。
英文部分：
Bhansali, A., 2016, “Forecasting the Future Through Machine Learning”, Microsoft Machine Learning Blog.
Chang, S. C., Tang, Y. C. and Chu, J. K., 2014, “Revisiting Online Repeat Purchase Behavior: Deterministic Versus Stochastic Behavioral Modeling”, Journal of Data Analysis. (9:5), pp. 23-46.
Colombo, R., and Jiang, W., 1999, “A Stochastic RFM Model”, Journal of Interactive Marketing, Vol. 13, No. 3, pp. 2-12
Deng, D., 2016, “Predictive Analysis for Telecom-Integrating Azure Data Lake and Azure ML”, Microsoft Machine Learning Blog.
Efron, B. and Tibshirani, R. J., 1994, “An Introduction to the Bootstrap”, Chapman & Hall/CRC, United Kingdom.
Gupta, S. and Lehmann D. R., 2003, “Customers As Assets”, Journal of Interactive Marketing, Vol. 17, No. 1, pp. 9-24.
He, H. and Garcia, E. A., 2008, “Learning from Imbalanced Data”, IEEE Transactions on Knowledge and Data Engineering (TKDE), Vol. 21, No. 9, September 2009.
Hsiao, B., Shu, L. and Hsieh, S. J., 2015a, “Predicting and Testing Probability of Continuous Purchasing of Online Customers”, 12th International Conference on Fuzzy Systems and Knowledge Discovery (FSKD).
Hsiao, B., Shu, L., Hsieh, S. J. and Chen, P. Y., 2015b, “Trial Plan With Capitation Payment of the National Healthcare Insurance in Taiwan: Establishing a Loyal Patient Selection Model”, PACIS 2015 Proceedings. 245.
Oshri, G., 2016, “RFM: A Simple and Powerful Approach to Event Modeling”, Microsoft Machine Learning Blog.
(https://blogs.technet.microsoft.com/machinelearning/2016/05/31/rfm-a-simpleand-powerful-approach-to-event-modeling/)
Provost, F. and Fawcett, T., 2013, “Data Science for Business”, Oreilly & Associates Inc, America.
Schmittlein, D. C., Morrison, D. G. and Cowmbo, R., 1987, “Counting your customers who are they and what will they do next”, Management science (33:1), pp. 1-24.
Schmittlein, D. C. and Peterson, R. A., 1994, “Customer base analysis: An industrial purchase process application”, Management science (13:1), pp. 41-67.
Wang, 2000. “The Relationships among Consumer Attributes, Users’ Satisfactions on Web and WWW Purchase Intention”, JBA(48), pp. 121-137.