在連鎖生鮮超市營運中，由於寄送實體的直郵廣告(Direct Mail, DM)預算有限，為了讓超市行銷人員可以有效的針對高消費能力，且具有較高機率會回超市消費的顧客群進行實體DM的寄送，我們依據概念飄移(Concept drift)的觀點，將『時間函數』與『消費過去行為』兩個因素結合，針對每位顧客建立其回購力指數(Re-purchase Index with Time factor, RIT) ；當顧客回購力指數愈高表示該顧客不但消費能力高且會回來消費的機率 (回購率)也高。
本論文中使用國內知名的連鎖生鮮超市所提供之交易資料，針對每位會員建立其回購力指數，我們將利用回購力指數針對不同消費行為之顧客，找尋寄送DM的目標客戶。最後，本論文將對目前生鮮超市只對總消費金額的高客戶群寄送實體DM的做法，與本文提出的回購力指數方法進行比較分析；經由實驗數據顯示，在找尋寄送DM的目標客戶時，利用回購力指數的方法不但能可以更有效的選取高回購力顧客，還可以降低選取到相同顧客的機率(重複率)；在有限的預算下，能夠選出更多不同的顧客，更有效的針對具消費力的潛在顧客寄送DM。

While operating supermarket businesses, we would like to target customer for marketers to mail direct mails(DM) because of the limitations of budget. Beside, high repurchase rate didn't mean that customers had high purchase ability, high purchase ability might not present that customers have high repurchase rate. To let marketers mailing DMs to the right target customers, in this paper, we would discuss the relations between purchase behavior and repurchase rate and use transaction data of customers to construct Re-purchase Index with Time factor (RIT) for every customer. Then, we would predict the repurchase rate of customers that have high purchase capacity with concept drift and purchase behavior.
Due to the current supermarket marketing target customers were chosen based on the total money that they had spent. In this article, we would analyze and question this model. At the same time, we would like to target customers that had different purchase behavior based on RIT model, so that, marketers could select the best target customers with higher possibility revisit rate, to reach the goal of increasing profit by reducing the costs of mailing DMs for supermarket.

目錄
第1章 緒論 1
1.1 研究背景與動機 1
1.2 研究架構 6
第2章 相關文獻與研究探討 7
2.1 顧客輪廓(Customer profile) 8
2.2 概念飄移(Concept drift) 10
第3章 問題陳述 16
3.1 回購力指數(Re-Purchase ability Index, RI)的問題 17
3.2 考慮時間因素之回購力指數(RIT) 22
第4章 研究方法與驗證 32
4.1 驗證RIT模型的預測回購顧客與實際回購率之比較 33
4.2 RI模型與RIT模型的比較 40
第5章 實驗結果與探討 44
5.1 某知名超市選取顧客之做法與RIT模型比較 45
5.2 考慮到實際行銷預算有限的顧客選取 56
第6章 結論與未來研究方向 68
參考文獻 69
附錄-英文論文 72

圖目錄
圖1 1至9月所有顧客在10是否有消費人數比例 3
圖2 最後一次交易時間距離9/30的間隔天數與人數比及10月實際到店人數 4
圖3 1到9月曾消費過之顧客在10月之實際回購率 5
圖4 3種不同的參數所呈現的時間函數曲線 13
圖5 THE ALGORITHM OF THE CPIT MODEL 15
圖6 以5月RI預測6月實際回購率 18
圖7 6月實際消費金額 19
圖8 為該年4月到6月的顧客前3個月回購情形 22
圖9 為5月有來消費客戶前3個月的消費情況 23
圖10 4到6月前12周到前1周回購率 24
圖11 4月的預測回購率與實際回購率比較 28
圖12 5月的預測回購率與實際回購率比較 28
圖13 6月的預測回購率與實際回購率比較 29
圖14 4月的實際平均消費金額 30
圖15 5月的實際平均消費金額 30
圖16 6月的實際平均消費金額 31
圖17 7月的預測與實際回購率 33
圖18 8月的預測與實際回購率 34
圖19 9月的預測與實際回購率 34
圖20 10月的預測與實際回購率 35
圖21 11月的預測與實際回購率 35
圖22 7月實際消費總金額 37
圖23 8月實際消費總金額 37
圖24 9月實際消費總金額 38
圖25 10月實際消費總金額 38
圖26 11月實際消費總金額 39
圖27 RIT6與RI6在7月實際回購率之比較 41
圖28 RIT7與RI7在8月實際回購率之比較 41
圖29 RIT8與RI8在9月實際回購率之比較 42
圖30 RIT9與RI9在10月實際回購率之比較 42
圖31 RIT10與RI10在11月實際回購率之比較 43
圖32 RIT模型與超市做法(依消費能力高低)在7月實際回購率之比較 46
圖33 RIT模型與超市做法(依消費能力高低)在8月實際回購率之比較 46
圖34 RIT模型與超市做法(依消費能力高低)在9月實際回購率之比較 47
圖35 RIT模型與超市做法(依消費能力高低)在10月實際回購率之比較 47
圖36 RIT模型與超市做法(依消費能力高低)在11月實際回購率之比較 48
圖37 RIT模型與超市做法(依消費能力高低)在7月的平均消費能力之比較 49
圖38 RIT模型與超市做法(依消費能力高低)在8月的平均消費能力之比較 49
圖39 RIT模型與超市做法(依消費能力高低)在9月的平均消費能力之比較 50
圖40 RIT模型與超市做法(依消費能力高低)在10月的平均消費能力之比較 50
圖41 RIT模型與超市做法(依消費能力高低)在11月的平均消費能力之比較 51
圖42 RIT模型與超市做法(依消費能力高低)在7月的總消費力之比較 52
圖43 RIT模型與超市做法(依消費能力高低)在8月的總消費力之比較 52
圖44 RIT模型與超市做法(依消費能力高低)在9月的總消費力之比較 53
圖45 RIT模型與超市做法(依消費能力高低)在10月的總消費力之比較 53
圖46 RIT模型與超市做法(依消費能力高低)在11月的總消費力之比較 54
圖47 在7到11月選取500人的實際回購率 56
圖48 在7到11月選取1000人的實際回購率 57
圖49 在7到11月選取1500人的實際回購率 57
圖50 在7到11月選取2000人的實際回購率 58
圖51 在7到11月選取500人的顧客選取重複率 59
圖52 在7到11月選取1000人的顧客選取重複率 60
圖53 在7到11月選取1500人的顧客選取重複率 60
圖54 在7到11月選取2000人的顧客選取重複率 61
圖55 選取即將流失的500名顧客之兩種做法比較 62
圖56 選取即將流失的1000名顧客之兩種做法比較 63
圖57 選取即將流失的1500名顧客之兩種做法比較 63
圖58 選取即將流失的2000名顧客之兩種做法比較 64
圖59 在7到11月中，選取即將流失的顧客500人之重複率 66
圖60 在7到11月中，選取即將流失的顧客1000人之重複率 66
圖61 在7到11月中，選取即將流失的顧客1500人之重複率 67
圖62 在7到11月中，選取即將流失的顧客2000人之重複率 67
 
表目錄
表1 RI_5第3群中，不同類型之消費者的回購率 20
表2 為RI_5為0不同類型之顧客在6月的回購率 20
表3 為兩種不同的消費者回購率之比例 26

[1]	Y. Chen, D. Pavlov, and J. F. Canny, "Behavioral Targeting: The Art of Scaling Up Simple Algorithms," ACM Trans. Knowl. Discov. Data, vol. 4, pp. 1-31, 2010.
[2]	D. Peppers and M. Rogers, Return on customer: Creating maximum value from your scarcest resource: Doubleday/Currency, 2005.
[3]	A. Tsymbal, "The problem of concept drift: definitions and related work," Computer Science Department, Trinity College Dublin, 2004.
[4]	R. M. Bell, Y. Koren, and C. Volinsky, "The bellkor 2008 solution to the netflix prize," Statistics Research Department at AT&T Research, 2008.
[5]	L. Xiang and Q. Yang, "Time-dependent models in collaborative filtering based recommender system," in Web Intelligence and Intelligent Agent Technologies, 2009. WI-IAT'09. IEEE/WIC/ACM International Joint Conferences on, 2009, pp. 450-457.
[6]	Y. Koren, "Collaborative filtering with temporal dynamics," Communications of the ACM, vol. 53, pp. 89-97, 2010.
[7]	Y. Ding and X. Li, "Time weight collaborative filtering," in Proceedings of the 14th ACM international conference on Information and knowledge management, 2005, pp. 485-492.
[8]	S. Ma, X. Li, Y. Ding, and M. E. Orlowska, "A recommender system with interest-drifting," in Web Information Systems Engineering–WISE 2007, ed: Springer, 2007, pp. 633-642.
[9]	J. Zhao, X. Yu, and J. Sun, "TDCF: time distribution collaborative filtering algorithm," in Proceedings of the 2008 International Symposium on Information Science and Engieering-Volume 01, 2008, pp. 98-101.
[10]	Y. Blanco-Fernandez, J. J. Pazos-Arias, M. Lopez-Nores, A. Gil-Solla, M. Ramos-Cabrer, J. Garcia-Duque, et al., "Incentivized provision of metadata, semantic reasoning and time-driven filtering: Making a puzzle of personalized e-commerce," Expert Systems with Applications, vol. 37, pp. 61-69, 2010.
[11]	W. Fan, "Systematic data selection to mine concept-drifting data streams," in Proceedings of the tenth ACM SIGKDD international conference on Knowledge discovery and data mining, 2004, pp. 128-137.
[12]	R.-D. Chiang, Y.-H. Wang, and H.-C. Chu, "Prediction of members’ return visit rates using a time factor," Electronic Commerce Research and Applications, vol. 12, pp. 362-371, 9// 2013.
[13]	M. Balabanović and Y. Shoham, "Fab: content-based, collaborative recommendation," Communications of the ACM, vol. 40, pp. 66-72, 1997.
[14]	M. J. Pazzani, J. Muramatsu, and D. Billsus, "Syskill & Webert: Identifying interesting web sites," in AAAI/IAAI, Vol. 1, 1996, pp. 54-61.
[15]	J. Shavlik, S. Calcari, T. Eliassi-Rad, and J. Solock, "An instructable, adaptive interface for discovering and monitoring information on the World-Wide Web," in Proceedings of the 4th international conference on Intelligent user interfaces, 1998, pp. 157-160.
[16]	D. Kelly and J. Teevan, "Implicit feedback for inferring user preference: a bibliography," in ACM SIGIR Forum, 2003, pp. 18-28.
[17]	B. Mobasher, "Data mining for web personalization," in The adaptive web, ed: Springer, 2007, pp. 90-135.
[18]	J. Akehurst, I. Koprinska, K. Yacef, L. Pizzato, J. Kay, and T. Rej, "Explicit and implicit user preferences in online dating," in New Frontiers in Applied Data Mining, ed: Springer, 2012, pp. 15-27.
[19]	E. R. Nunez-Valdez, J. M. Cueva Lovelle, O. Sanjuan Martinez, V. Garcia-Diaz, P. Ordonez de Pablos, and C. E. Montenegro Marin, "Implicit feedback techniques on recommender systems applied to electronic books," Computers in Human Behavior, vol. 28, pp. 1186-1193, 2012.
[20]	R. Cooley, B. Mobasher, and J. Srivastava, "Data preparation for mining world wide web browsing patterns," Knowledge and information systems, vol. 1, pp. 5-32, 1999.
[21]	S.-H. Min and I. Han, "Detection of the customer time-variant pattern for improving recommender systems," Expert Systems with Applications, vol. 28, pp. 189-199, 2005.
[22]	J. Lee, M. Podlaseck, E. Schonberg, and R. Hoch, "Visualization and analysis of clickstream data of online stores for understanding web merchandising," in Applications of Data Mining to Electronic Commerce, ed: Springer, 2001, pp. 59-84.
[23]	Y. H. Cho and J. K. Kim, "Application of Web usage mining and product taxonomy to collaborative recommendations in e-commerce," Expert systems with Applications, vol. 26, pp. 233-246, 2004.
[24]	J. Cho, K. Kwon, and Y. Park, "Collaborative filtering using dual information sources," Intelligent Systems, IEEE, vol. 22, pp. 30-38, 2007.
[25]	A. Albadvi and M. Shahbazi, "A hybrid recommendation technique based on product category attributes," Expert Systems with Applications, vol. 36, pp. 11480-11488, 2009.
[26]	Y.-J. Park and K.-N. Chang, "Individual and group behavior-based customer profile model for personalized product recommendation," Expert Systems with Applications, vol. 36, pp. 1932-1939, 2009.
[27]	P. Cunningham, N. Nowlan, S. J. Delany, and M. Haahr, "A case-based approach to spam filtering that can track concept drift," in The ICCBR, 2003, pp. 03-2003.
[28]	M. Salganicoff, "Tolerating concept and sampling shift in lazy learning using prediction error context switching," Artificial Intelligence Review, vol. 11, pp. 133-155, 1997.