1 / 46

Agenda

In the dynamic retail market, understanding changes in customer behavior can ... Case Study(Application of dynamic CRM to a retailer) Paper 2 Framework of analysis ...

liam
Download Presentation

Agenda

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


    Using Data Mining Technology to Evaluate Customer’s Time-Variant Purchase Behavior and Corresponding Marketing Strategies ????: ???? ?? ???? ?? ??: ??? 923834 OR seminar

    Slide 2:Agenda

    Background Introduction Approach Conclusion

    Slide 3:In the past, researchers generally applied statistical surveys to study customer behavior. Recently, however, data mining techniques have been adopted to predict customer behavior.

    Background ???????,?????????????????CRM?????(buying-behavior-based CRM) , ????CRM??????,????????data mining?????? ???????,?????????????????CRM?????(buying-behavior-based CRM) , ????CRM??????,????????data mining??????

    Slide 4:????????????

    Technology Review??(??????2002?1???)????????????? ???????? ????? ???? (Data mining) ?????? ????? (Biometrics) ?????? ????? (Microphotonics) ????? (Untangling code) ????? ?????? (Microfluidics)

    Slide 5:Data mining???????????????????,???Berry?Linoff (1997) ?????,?????????????????????????????????????,??? (Trend)???(Pattern)???? (Relationship)? ??????????:?? (classification)??? (estimation)??? (prediction)????? (affinity grouping)???(clustering) ?

    Classification Estimation Prediction Association rule Clustering Data mining Data mining

    Slide 6:The Goal of CRM (Customer Relation Management)

    Identify the customer ????????,??????????,?????????,????????????,???????,???????????????Internet???????????,?????????????? Construct customer purchase data mart Understand and predict the customer-buying pattern ?????????????????????????,???????

    Slide 7:RFM (Recently?Frequency?Monetary) ,???????????????????,??????????????????????? ??????(R): ???????????????????????????????,???????????????,??????????????,???????,??????????????? ????(F): ?????????????????????????????????????,????????????????????? ????(M): ??????????????,????????????????????????,?????????????

    Measure purchase behavior of customer

    Slide 8:Kahan (1998) ??RFM??????????????,??RFM????????????????,???????????,RFM?????????????????? ?RFM????,Sung and Sang (1998)?RFM?????????????,????????????????????,?????????????????

    Measure purchase behavior of customer

    Slide 9:Introduction

    In the dynamic retail market, understanding changes in customer behavior can help managers to establish effective promotion campaigns. The proposed approach can assist managers in developing better marketing strategies.

    Slide 10:Topic

    Chen, M.C., Chiu, A.L., and Chang, H.H., 2005, “Mining changes in customer behavior in retail marketing”, Expert Systems With Applications, 28(4), pp. 773-781. Sung H.H., Sung M.B., and Sang C.P., 2002, “Customer’s time-variant purchase behavior and corresponding marketing strategies: an online retailer’s case”, Computer & Industrial Engineering, 43, pp. 801-820. Time-variant Behavior Analysis - Markov Chain Time-variant Behavior Analysis - Association rule mining

    Slide 11:???paper????CRM???,??????????????,????CRM Model??Data Mining ? Monitoring Agent System (MAS)??????????????? ???paper???? Customer Behavioral Variables, Demographic Variables, ? Transaction Database ???????,?????????Data Mining???????????????

    Slide 12:Agenda

    Background Introduction Approach Paper 1 Framework of analysis Monitoring Agent System Dynamic CRM Model Case Study(Application of dynamic CRM to a retailer) Paper 2 Framework of analysis Conclusion

    Slide 13:Monitoring Agent System(1/2)

    Slide 14:Monitoring Agent System(2/2)

    ????????????????,??Monitoring Agent System (MAS) ?????????????????????????,????????????????,?????????,??????: 1. ?Customer purchase data mart?????????? 2. ???????summarize??????????,R?F??M? 3-1. ?????Customer career path DB? 3-2. ?????segmentation? 3-3. ?????RFM??????? 4. ?segmentation??????????RFM???????,??R?F?M?(????R?F?M????????,??????)?R?F?M??R?F?M??R?F?M??R?F?M??R?F?M??R?F?M? ? R?F?M?? 5.. ?????????Customer segment knowledge DB?? 6. ??Customer segment knowledge DB?????,??????????????? 7.. ???????,?????,?????;???????,?????????

    Slide 15:Dynamic CRM Model (1/9)

    Assumed: the model has the Markovian property The process will evolve in the future depend only on the present state of the process. A special kind of stochastic process.

    Slide 16:Dynamic CRM Model (2/9)

    States Transition Probability Matrix ?????n-1?????m,???n?????k??????????????

    Slide 17:Dynamic CRM Model (3/9)

    Example ?? 4??? ?? 4???

    Slide 18:Dynamic CRM Model (4/9)

    ????????????,????????????,???A?B??C,???????????? ,??????????????100,????t?t+1,??????????,???????????????? ??????????????,??????????? ,Markov Chain?????????,?????????????????????????????????,?????????? Example

    Slide 19:Dynamic CRM Model (5/9)

    Example (Cont.) Hypothetical Profit Rate: Segment A=15%, Segment B=25%, Segment C=40% Original: After Promotion:

    Slide 20:Dynamic CRM Model (6/9)

    Evaluating alternative marketing strategies Original: Strage1 Strage2

    Slide 21:Dynamic CRM Model (7/9)

    ???????????????,?????????,??????????,??????????? ? ??A?????????????,???C??????????? ? ??????????,???????,??A?B????,?60%?????????C,???C????50%?VIP ??? ?????,????B????????,????C??????????????,???????????

    Slide 22:Dynamic CRM Model (8/9)

    Evaluating alternative marketing strategies Monitoring the movements of segments

    Slide 23:Dynamic CRM Model (10/10)

    Assumption Relaxation Have New Customer Have Defector Customer

    Slide 24:Case Study

    ????????? (RFM) ??,???1995??????1996?12?31?,?????2036?????? ???????? (???????????),??????RFM? ?

    Slide 25:Rt: measures how long it has been since he or she made a last purchase during last observation period from time t. Ft: measures how many times he or she has purchased products during that period. Mt: measures how much he or she has spent in total.

    R???????????,??????????????,?????R(??????????)???,??????????,?R???????????????R?? F???????????,??????????????,F???????F?,??,?????????,?F??????????????????F?? M??????????,?????????????,?M???????M?,??,?????????,?M????????????????????M?? R???????????,??????????????,?????R(??????????)???,??????????,?R???????????????R?? F???????????,??????????????,F???????F?,??,?????????,?F??????????????????F?? M??????????,?????????????,?M???????M?,??,?????????,?M????????????????????M??

    Slide 26:Customer Clustering

    SOM Training the SOM Mapping input customer RFM patterns to output customer segments Label of segments If each average of segments is bigger than the overall mean, a character ‘h’ is given to that value. If the opposite case occurs, a character ‘l’ is given. ?????????????RFM??,?????????,??????????SOM???? ????????????,?RFM??SOM???,????????,???????,??????????????,???????????????? ???????,???????????,??????,??????????????A??????????????,????????A(h),????,????A(l),?????????????, ?????????????RFM??,?????????,??????????SOM???? ????????????,?RFM??SOM???,????????,???????,??????????????,???????????????? ???????,???????????,??????,??????????????A??????????????,????????A(h),????,????A(l),?????????????,

    Slide 27:Customer Segments and Corresponding Marketing Strategies at a Specific Time

    ????????????????????: 1.RhFlMl?????,??????,??????????,??????,???????????????????? 2.RlFlMl?????,?????,????????????????,???????????? 3.RlFhMh?????,?????,??????????????????,????????????,???????????????

    Slide 28:Changes of The Number of Customers in Each Segment

    ??????????,???????????,??44??,????????????,?????????????

    Slide 29:The Matrix of Transition Probability

    ????????????,??????????????RhFlMl????(0.05,27?)???RhFlMl?????????(0.062+0.115,66?)??????????RlFhMh????(0.062+0.171,82?)???RlFhMh???????????(0.056,16?)? ??????????????????,???????????????,?????????????? ????? ???? ???? ????? ???? ????

    Slide 30:Agenda

    Background Introduction Approach Paper 1 Framework of analysis Paper 2 Framework of analysis Data pre-processing and Customer segmentation Association rule mining Change patterns Case Result Conclusion

    Slide 31:Flowchart of mining changes for customer behavior

    1. ??,??,????????????ER Model????data (RFM???????) integration and transformation? 2. ????association rules ?mining ????,?????????(t1,t2)?datasets???association rules ?????????? 3. ?????,???????????patterm ? 4. ?????patterm????????????????????

    Slide 32:Data pre-processing and Customer segmentation

    ??”????????”,?????????????(????????,????????)???????? ??F(????)?M(??????)??????, R(????)??????????????,???????????????? ?????????,???paper?????????4???????????????,???patterm ? ((

    Slide 33:Association rule mining

    ?????????????????????,?????????????????,???????????? Applications: ?????Walmart??????????????? ?????????????????,?????? ?????????? Examples: Rule form: “Body Head [support, confidence]”. buys(x, “diapers”) buys(x, “beers”) [0.5%, 60%] ????0.5%??????, 60%??????????? ?????? (Support:how useful is the rule) ???? (Confidence:how true) ????????, ???????????????????????????? (Support:how useful is the rule) ???? (Confidence:how true) ????????, ??????????????????????

    Slide 34:Change patterns(1/2)

    Emerging patterns:imply the same consumer behavior that exists in different periods of time with trend (the conditional and consequent parts are identical for and ). Added patterns:A rule at period t2, , is identified as an added pattern if all conditional and consequent parts differ significantly from any rule, , at period t1. The rule matching threshold (RMT) is used to measure the degree of change. (??,???,??,????) (??,???,??,????)

    Slide 35:Change patterns(2/2)

    Perished patterns:A rule at period t1, , is identified as a perished pattern if all conditional and consequent parts differ significantly from any rule, , at period t2. Unexpected changes: If the conditional parts of and are similar, but their consequent parts are different, then is an unexpected consequent change with respect to . If the consequent parts of and are similar, but their conditional parts are different, then is an unexpected conditional change with respect to . (??,???,??,????) (??,???,??,????)

    Slide 36:Change patterns

    To identify the degree of similarity and difference in customer behavior changes for different periods of time, this study designs two measures of similarity and unexpectedness. Similarity can be used to measure the degree of likeness between two rules. Unexpectedness can be used to identify the disparity between dissimilar rules. (??,???,??,????) (??,???,??,????)

    Slide 37:Change patterns

    Similarity:

    Slide 38:Examples: ????????? , ????: 1. Sex=F, 30days < Recency, Frequency > 13 times Buy =Vegetables, Size= Big, Color =Red 2. Sex=F, 10days < Recency < 20 days, Frequency > 13 times, 12< Monetary Buy =Vegetables, Size= Small

    Change patterns ? lij = 3 / 4, hij = 2 / 3 (?LHS or RHS ?????) ? Sij = 2 /4 * 1/3 = 1/6 (?LHS or RHS ??????????)

    Slide 39:Change patterns

    Maximum degrees of similarity: provides the basis for differentiating emerging patterns, added patterns, and perished patterns during various periods. (a) Emerging pattern: =1 (or =1) (b) Added pattern: < RMT; (c) Perished pattern: < RMT;

    Slide 40:Change patterns

    Unexpectedness : When the similarity between two rules equals 0, an unexpectedness measure is used to judge whether the two rules consist of unexpected changes. (d) Unexpected condition pattern: = -1 (<0) (e) Unexpected consequent pattern: = 1 (>1)

    Slide 41:Case Result

    Partial list of patterns for customer behavior.

    Slide 42:Case Result

    Emerging patterns: Added patterns:

    Slide 43:Case Result

    Perished patterns: Unexpected changes:

    Slide 44:Conclusion

    Mining changes for customer behavior is useful for satisfying customer needs in dynamic business environments. An online query system provides marketing managers a tool for marketing rapidly establish marketing strategies. Conclusion

    Slide 45:Reference

    Chen, M.C., Chiu, A.L., and Chang, H.H., 2005, “Mining changes in customer behavior in retail marketing”, Expert Systems With Applications, 28(4), pp. 773-781. Sung H.H., Sung M.B., and Sang C.P., 2002, “Customer’s time-variant purchase behavior and corresponding marketing strategies: an online retailer’s case”, Computer & Industrial Engineering, 43, pp. 801-820. Rygielski, C., Wang J.C., David Y.C., 2002, “Data mining techniques for customer relationship management”, Technology in Society, 24, pp. 483-502. Predicting Mail-Order Repeat Buying: Which Variables Matter? (http://ideas.repec.org/p/rug/rugwps/03-191.html) Berry, M., and Linoff, G., 1997, “Data mining techniques: for marketing, sales, and customer support”, John Wiley & Sons, Inc., NY. *

    The End Thanks!
More Related