IEEM 7103 Topics in Operations Research

1
IEEM 7103 Topics in Operations Research
Presentations

• Using Data Mining Technology to
  Evaluate Customers Time-Variant Purchase
  Behavior and Corresponding Marketing Strategies.

d923834 ???
2

• Paper
• Customers time-variant purchase behavior and
  corresponding marketing strategies an online
  retailers case
• Publish
• Computer Industrial Engineering 2002
• Authors
• Sung Ho Ha
• School of Business Administration, College of
  Economy Commerce, Kyungpook National
  University.
• Sung Min Bae, Sang Chan Park
• Department of Industrial Engineering, Korea
  Advanced Institute of Science and Technology.

3
Agenda

• Abstract
• Introduction
• Literature review
• Framework of analysis
• Application of dynamic CRM to a retailer (Case
  Study)
• Conclusion Future work

4
Abstract

• ?????CRM (Customer Relationship Management)
  ??????????????????????????????,????????????????,??
  ????
• ???????,???????? (buying-behavior-based
  CRM)?????????CRM???,?????CRM???????
• ??paper????CRM???,????????????????,????CRM
  Model??Data Mining ? Monitoring Agent System
  (MAS)???????????????

5
Agenda

• Abstract
• Introduction
• The Goal of CRM
• Applications
• The Lacks of Static CRM
• Literature review
• Framework of analysis
• Application of dynamic CRM to a retailer (Case
  Study)
• Conclusion Future work

6
Introduction

• The maturity of the business-to-consumer (B2C)
  market.
• A successful retailer must have provide a bundle
  of customized services.
• Consumer markets characteristics
• Repeat-buying over the relevant time horizon.
• A large number of customer.
• A wealth of information detailer past customer
  purchase.

7
The Goal of CRM

• Identify the customer
• Construct customer purchase data mart
• Understand and predict the customer-buying
  pattern

8
The Goal of CRM

• Measure purchase behavior of customer
• Recency????????????
• Frequency??????????????
• Monetary values??????????????
• ??????????????,???????

9
Applications

• CRM applications
• Short-range???????????,???????????????
• Intermediate-range???????????????????????
• Long-range?????????????????????
• EC (electronic commerce) applications
• ??????????????,???????????,?????????????,???
  CRM?????????

10
The Lacks of Static CRM

• ???????????????????????????
• ????????????????????
• ?????????????????????????
• ?????????????????????????

11
Agenda

• Abstract
• Literature review
• Framework of analysis
• Application of dynamic CRM to a retailer (Case
  Study)
• Conclusion Future work

12
Literature Review (1/3)

• Hughes, 1996
• CRM???????????????,??????????????????????????????
• Peppers, Rogers, Dorf, 1999
• ??????????????????,???????????????,????????????
• Peppard, 2000
• ??????????,????????????????,??????,?????????,?????
  ?????????
• Schafer, Konstan, Riedl, 2001
• ???web marketing,?????????????????????

13
Literature Review (2/3)

• Technique for online marketing
• Database marketing
• ?????????,??ZIP?income???????????segment as a
  group?
• ????????????????
• Ad targeting (offer targeting )
• ???????????,??????????????

14
Literature Review (3/3)

• One-to-one marketing (Peppard, Rogers, 1997a/b)
• ????????,???????,?????????????
• Content-based filtering system??????????,????????
  ???????
• Collaborative filtering system???????????,???????
  ???????????,??????

15
Agenda

• Abstract
• Literature review
• Framework of analysis
• Data Mining Analysis and Technique
• Monitoring Agent System
• Dynamic CRM Model
• Application of dynamic CRM to a retailer (Case
  Study)
• Conclusion Future work

16
Data Mining Analysis and Technique

• Time-variant Behavior Analysis
• Markov Chain
• Segmentation
• Self-Organization Map (SOM)
• Purchase Behavior Feature
• R (Recency), F (Frequency), M (Monetary)
• Classification
• Decision Tree (C4.5)

17
Monitoring Agent System
18
Dynamic CRM Model (1/10)

• Assumed the model has the Markovian property

?????????????CRM??,??????????????????????,????????
,?????????????????
A special kind of stochastic process.
The process will evolve in the future depend only
on the present state of the process.
19
Dynamic CRM Model (2/10)

• Assumed the model has the Markovian property
  (Cont.)

20
Dynamic CRM Model (3/10)

• States Transition Probability Matrix

21
Dynamic CRM Model (4/10)

• Example

22
Dynamic CRM Model (5/10)

• Stability of the matrix of transition
  probabilities
• Example

In the long run, the process usually approaches a
steady state or equilibrium when the systems
state probabilities have not changed further so
long as the matrix of transition probabilities
remains the same.
23
Dynamic CRM Model (6/10)

• Example (Cont.)
• Hypothetical Profit Rate Segment A15, Segment
  B25, Segment C40

Original
After Promotion
24
Dynamic CRM Model (7/10)

• Evaluating alternative marketing strategies

Original
Strage1
Strage2
25
Dynamic CRM Model (8/10)

• Evaluating alternative marketing strategies
  (Cont.)

26
Dynamic CRM Model (9/10)

• Monitoring the movements of segments

27
Dynamic CRM Model (10/10)

• Assumption Relaxation
• Have New Customer
• Have Defector Customer

28
Agenda

• Abstract
• Literature review
• Framework of analysis
• Application of dynamic CRM to a retailer (Case
  Study)
• Conclusion Future work

29
Case Study

• ????????? (RFM) ??,???1995??????1996?12?31?,?????2
  036??????
• ???????? (???????????),??????RFM? ?

30

• 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.

31
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.

32
Customer Segments and Corresponding Marketing
Strategies at a Specific Time
33
Changes of The Number of Customers in Each Segment
34
The Matrix of Transition Probability

• ????????????,??????????????RhFlMl????(0.05,27?)???
  RhFlMl?????????(0.0620.115,66?)??????????RlFhMh??
  ??(0.0620.171,82?)???RlFhMh???????????(0.056,16?)
  ?
• ??????????????????,???????????????,??????????????

????
?????
????
?????
????
????
35
Comparison of current strategy
36
Agenda

• Abstract
• Literature review
• Framework of analysis
• Application of dynamic CRM to a retailer (Case
  Study)
• Conclusion Future work

37
Conclusion Future Work (1/4)

• Conclusion
• A cost-effective method for application of CRM
  should be done dynamically in time to solve
  management problems.
• A method to discover potential customers.
• Future work
• The analysis of the outlier in loyal cluster.
• Extend one-dimensional features to
  multi-dimensional features.

38
Future Work (2/4)

• To use a multi-channel contact center as the
  foundation to model customer interaction
  processes.
• To develop contact center evaluation methods, and
  define key performance indicators (KPI) for
  continuously improving customer service quality.

39
Future Work (3/4)

• Contact Centers
• Contact centers are implemented to provide
  improved customer services (Rowley, Mostowfi, and
  Lees, 2002).
• Contact centers not only integrate multi-channels
  and provide customer service supports, but also
  improve quality management, contact routing, and
  knowledge management.

40
Future Work (4/4) - Multi-channels Contact Center

41
The End
42
Contact Center Clustering Analysis

• Sharma (1996) proposed the RMSSTD (Root Mean
  Square Standard Deviation) and RS (R-Squared)
  methods to evaluate the quality of
  non-hierarchical clustering (e.g. K-Means) result.

43
Key Performance Indicators

• Key Performance Indicators (KPIs) measure
  performance and improvements after companies have
  implemented new business processes.
• The goal of implementing KPIs is to create
  professional contact centers and deliver high
  levels of service quality.

44
Key Performance Indicators

• Cleveland (1996), Goodman, Ward, Segal and
  Cleveland (2000), Feinberg, Kim and Hokama (2000)
  and Grimm (2001) indicate the following important
  KPIs for contact center
• Average time of answering inquiries.
• Customer queuing time
• Percentage of callers who have satisfactory
  resolution of problems on the first call
• Abandonment rate (the percentage of callers who
  hang up or disconnect prior to answers being
  provided)
• Adherence (agents in their seats and working as
  scheduled)
• Average work time after call (time needed to
  finish paper work and research after the call has
  initially been handled)
• Percentage of calls blocked (percentage of
  callers who receive a busy signal and cannot get
  into the service queue)
• Time before abandoning the call (average time
  caller waited before balking from the service
  queue)
• Inbound calls per agent during the work shift
• Agent turnover (the number of agent who leave
  employment during a period of time)
• Total cost of delivering the contact center
  service
• Service output level (total number of customers
  served in a period of time)
• Revenue of the contact center
• The difference between predicted and actual work
  load
• The difference between predicted and actual agent
  demand
• Agent idle time