This document discusses techniques for customer relationship management (CRM) using data mining. It begins by introducing common data mining applications in retail, banking, and telecommunications. It then discusses how data mining can be used throughout the customer lifecycle to perform tasks like up-selling, cross-selling, and customer retention. The document proceeds to explain various data mining techniques including descriptive techniques like clustering and association rule mining as well as predictive techniques like classification, regression, and decision trees. It concludes by discussing major issues in the field of data mining.

1. Data Mining
Techniques for CRM
B.SHYAAMINI
(11MCM043)

2. 2
“The key in business is to know something that nobody
else knows.”
— Aristotle Onassis
“To understand is to perceive patterns.”
— Sir Isaiah Berlin
Data Mining Motivation

3. 3
Data Mining Applications

4. 4
Data Mining Applications:
RETAIL:
 Performing basket analysis
 Sales forecasting
 Database marketing
 Merchandise planning and allocation
BANKING:
 Fraud detection
 Card marketing
TELECOMMUNICATION:
 Call detail record analysis
 Customer loyalty

5. CRM Life Cycle:
5

6. 6
Data Mining in CRM:
Customer Life Cycle
 Customer Life Cycle
 The stages in the relationship between a customer
and a business
 Key stages in the customer lifecycle
 Prospects: people who are not yet customers but are
in the target market
 Responders: prospects who show an interest in a
product or service
 Active Customers: people who are currently using
the product or service
 Former Customers: may be “bad” customers who
did not pay their bills or who incurred high costs
 It’s important to know life cycle events (e.g. retirement)

7. 7
Data Mining in CRM:
Customer Life Cycle
 What marketers want: Increasing customer revenue and
customer profitability
 Up-sell
 Cross-sell
 Keeping the customers for a longer period of time
 Solution: Applying data mining

8. 8
Data Mining in CRM
 Data Mining helps to
 Determine the behavior surrounding a particular
lifecycle event
 Find other people in similar life stages and determine
which customers are following similar behavior
patterns

9. 9
Data Mining in CRM (cont.)
Data Warehouse Data Mining
Campaign Management
Customer Profile
Customer Life Cycle Info.

10. 10
Data Mining Techniques
Data Mining Techniques
Descriptive Predictive
Clustering
Association
Classification
Regression
Sequential Analysis
Decision Tree
Rule Induction
Neural Networks
Nearest Neighbor Classification

11. 11
Predictive Data Mining
Tridas Vickie Mike
Honest
BarneyWaldoWally
Crooked

12. 12
Prediction
Tridas Vickie Mike
Honest = has round eyes and a smile

13. 13
Decision Trees
 Data
height hair eyes class
short blond blue A
tall blond brown B
tall red blue A
short dark blue B
tall dark blue B
tall blond blue A
tall dark brown B
short blond brown B

14. 14
Decision Trees (cont.)
hair
dark
red
blond
short, blue = B
tall, blue = B
tall, brown= B
{tall, blue = A} short, blue = A
tall, brown = B
tall, blue = A
short, brown = B
Completely classifies dark-haired
and red-haired people
Does not completely classify
blonde-haired people.
More work is required

15. 15
Decision Trees (cont.)
hair
dark
red
blond
short, blue = B
tall, blue = B
tall, brown= B
{tall, blue = A} short, blue = A
tall, brown = B
tall, blue = A
short, brown = B
eye
blue brown
short = A
tall = A
tall = B
short = B
Decision tree is complete because
1. All 8 cases appear at nodes
2. At each node, all cases are in
the same class (A or B)

16. 16
Decision Trees:
Learned Predictive Rules
hair
eyesB
B
A
A
dark
red
blond
blue brown

17. 17
Decision Trees:
Another Example
Total list
50% member
0-1 child 2-3 child
20% member
4+ children
$50-75k income
15% member
$75k+ income
70% member
$50-75k income $20-50k income
85% member
Age: 40-60
80% member
Age: 20-40
45% member

18. 18
Rule Induction
 Try to find rules of the form
IF <left-hand-side> THEN <right-hand-side>
 This is the reverse of a rule-based agent, where the rules
are given and the agent must act. Here the actions are
given and we have to discover the rules!
 Prevalence = probability that LHS and RHS occur together
(sometimes called “support factor,” “leverage” or “lift”)
 Predictability = probability of RHS given LHS (sometimes
called “confidence” or “strength”)

19. 19
Association Rules from
Market Basket Analysis
 <Dairy-Milk-Refrigerated> → <Soft Drinks
Carbonated>
 prevalence = 4.99%, predictability = 22.89%
 <Dry Dinners - Pasta> → <Soup-Canned>
 prevalence = 0.94%, predictability = 28.14%
 <Dry Dinners - Pasta> → <Cereal - Ready to Eat>
 prevalence = 1.36%, predictability = 41.02%
 <Cheese Slices > → <Cereal - Ready to Eat>
 prevalence = 1.16%, predictability = 38.01%

20. 20
Use of Rule Associations
 Coupons, discounts
 Don’t give discounts on 2 items that are frequently
bought together. Use the discount on 1 to “pull”
the other
 Product placement
 Offer correlated products to the customer at the
same time. Increases sales
 Timing of cross-marketing
 Send camcorder offer to VCR purchasers 2-3
months after VCR purchase
 Discovery of patterns
 People who bought X, Y and Z (but not any pair)
bought W over half the time

21. 21
Finding Rule Associations
Algorithm
 Example: grocery shopping
 For each item, count # of occurrences (say out of 100,000)
 apples 1891, caviar 3, ice cream 1088, …
 Drop the ones that are below a minimum support level
 apples 1891, ice cream 1088, pet food 2451, …
 Make a table of each item against each other item:
 Discard cells below support threshold. Now make a cube for
triples, etc. Add 1 dimension for each product on LHS.
apples ice cream pet food
apples 1891 685 24
ice cream ----- 1088 322
pet food ----- ----- 2451

22. 22
Clustering
 The art of finding groups in data
 Objective: gather items from a database into sets
according to (unknown) common characteristics
 Much more difficult than classification since the classes
are not known in advance (no training)
 Technique: unsupervised learning

23. 23
The K-Means Clustering
Method
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5 6 7 8 9 10
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5 6 7 8 9 10
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5 6 7 8 9 10
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5 6 7 8 9 10
0
1
2
3
4
5
6
7
8
9
10
0 1 2 3 4 5 6 7 8 9 10
K=2
Arbitrarily choose K
objects as initial
cluster center
Assign
each of
the
objects
to most
similar
center
Update
the
cluster
means
Update
the
cluster
means
reassignreassign

24. Major Issues in Data
Mining
 Mining methodology and user interaction
 Mining different kinds of knowledge in databases
 Interactive mining of knowledge at multiple levels of
abstraction
 Incorporation of background knowledge
 Expression and visualization of data mining results
 Handling noise and incomplete data
 Pattern evaluation: the interestingness problem
 Performance and scalability
 Efficiency and scalability of data mining algorithms
 Parallel, distributed and incremental mining methods
24

25. Major Issues in Data
Mining
 Issues relating to the diversity of data types
 Handling relational and complex types of data
 Mining information from heterogeneous databases and
global information systems (WWW)
 Issues related to applications and social impacts
 Application of discovered knowledge
 Domain-specific data mining tools
 Intelligent query answering
 Process control and decision making
 Integration of the discovered knowledge with existing
knowledge: A knowledge fusion problem
 Protection of data security, integrity, and privacy
25

26. THANK YOU
26