Download Case Studies

Data Mining
Techniques for CRM
Paul J.C. Chang
Eneida Lau
Ximena Salazar
Lester Arellano
José-Pablo González
Edith Quispe
Data Mining in CRM ...
“ ...through data mining – the extraction of hidden
predictive information from large databases –
organizations can identify valuable customers, predict
future behaviors, and enable firms to make proactive,
knowledge-driven decisions.”
Agenda
 Introduction, Definition: Paul
 The Evolution & Apps. of Data Mining: Eneida
 Internal Considerations & Data mining techniques: Ximena
 Data mining and CRM – relationship & customer privacy: Lester
 Case Studies (Neural Networks, CHAID): JPG
 CHAID vs neural nets; Conclusions: Edith
Introduction
 Product-oriented view VS. Customer-oriented view
 Design-build-sell VS. sell-build-redesign
 One-on-one marketing VS. mass marketing
 Goal of revolution: Establish a long term
relationship with each customer
 The advent of the Internet and technological tools
accelerate modern CRM revolution
 CRM is important for B2C or C2B, and even more
crucial in B2B environments
Why Data Mining?
 Between businesses and customers…
 Collecting customer demographics and behavior data
makes precision targeting possible
 Helps to devise an effective promotion plan when
new products developed
 Creates and solidifies close customer relationships
Between businesses…
 Helps to smooth transactions, communications and
collaboration
 Simplifies and improves logistics and procurement
process
What is Data Mining?
 “…a sophisticated data search capability that uses
statistical algorithms to discover patterns and correlations
in data.”
 “…another way to find meaning in data.”
 Data mining is part of a larger process called knowledge
discovery
What Data Mining is ~NOT~
• Data mining software does not eliminate the
need to know the business, understand the
data, or be aware of general statistical
methods.
• DM does not find patterns or knowledge
without verification
• DM helps to generate hypotheses, but it does
not validate the hypotheses
Evolutionary Stages of Data Mining
Data
Collection
(1960’s)
Data
Access
(1980’s)
Data
Navigation
(1990’s)
Data
Mining
(2000’s)
•Retrospective,
static data delivery
•Retrospective,
dynamic data delivery
at record level
•Retrospective,
dynamic data delivery
at multiple level
•Retrospective,
Proactive information
delivery
•Summations or
averages
•Branch sales at
specific period of time
•Global view or drill
down
•Online analytic tools,
feedback and
information exchange
•RDBMS, SQL, ODBC
•OLAP,
multidimensional
databases, data
warehouses
•Computers, tapes,
disks
•IBM, CDC
•Oracle, Sybase,
Informix, IBM,
Microsoft
•Pilot, IRI, Arbor,
Redbrick
•Adv. Algorithms,
multiprocessor,
computers, massive
databases
•Lockheed, IBM, SGI
Breakdown of Data Mining from a Process
Orientation
Data Mining
Discovery
•Conditional
Logic
•Affinities and
Associations
•Trends and
Variations
Predictive
Modeling
Forensic
Analysis
•Outcome
Prediction
•Deviation
Detection
•Forecasting
•Link Analysis
Applications of Data Mining
Retail
1. Performing
basket analysis
2. Sales forecasting
3. Database
marketing
4. Merchandise
planning and
allocation
Banking
1. Card marketing
2. Cardholder
pricing and
profitability
3. Fraud detection
4. Predictive lifecycle
management
Telecommunications
1. Call detail record
analysis
2. Customer loyalty
OTHER APPLICATIONS
Customer
Segmentation
Discrete
segments by
adding variables
Manufacturing
Customize
Products.
Predict features
Warranties
No. clients who
will ask for claims
Frequent flier
incentives
Identify groups
who can receive
incentives
INTERNAL CONSIDERATIONS
Data mining
Decision-making process
Skillsets and technologies must be available to integrate them
Knowledge
gained
through DM
•
•
•
•
Sell to and service customers
Manage inventory
Supervise employees
Work to correct and prevent loss
-An algorithm for scoring
-A score for particular customer,
employee
-An action associated with a customer,
employee or transaction
DATA MINING TECHNIQUES
Nearest
Neighbor
Data Retained
DM
Approaches
Data distilled
Case-Based
Reasoning
Logical
Numeric and
Non-numeric
Cross
Tabulational
Non-numeric
Data
Equational
They are applied to tasks of predictive
modeling
and forensic analysis
They extract patterns and then use for various purposes
Numeric
Data
CUSTOMER RELATION MANAGEMENT
Definition
•
•
•
•
Know
Target
Sell
Service
1
2 Stage
Concept
2
CRM: Development
of the offer
3 Which’s
- From product to customer orientation
- Market Strategy from outside-in
-Push the development of customer orientation
-Innovating value proposition
Components of CRM
Customer
Information
Data
Warehouse
Analyze the
Data
Campaign
Execution &
Tracking
Internal
Customer
Data
Customer
Outside
Data
Source Data
Historical
Data
•Billing
Records
•Surveys
•Web logs,
•External
Credit Card
data
sources
records
Current
Address, Web
page viewing
profiles.
Data Mining Techniques +
Customer Oriented
Interactions between
MKT, information,
Tech and sales
channels
Data Mining & CRM
• The Relationship
– Customer Life
Cycle
• Prospects
• Respondents
• Active Customers
• Former Customers
Data
Mining
Inputs
What
information is
available
Output
What is likely
to be
interested
Data Mining & CRM
• Inputs
– Prospects Data Warehouse in other industries
– Click Stream Information
• Market Data Intelligence
– DM can predict behavior of customer (CLC) and match it with
any market event (a,i. I pod nano)
• Data Mining and Customer Privacy
– Privacy Bill of Rights, Independent verification of security
policies.
– Create an anonymous architecture for handling customer info.
Case Studies
Neural Networks vs. CHAID
Case #1
Neural Networks
Neural Networks
• The exact way in which
the brain enables
thought is one of the
great mysteries of
science
Neurons
NeoVistas Solutions’ Decision Series
• For retail, insurance, telecommunications, and
healthcare.
• Includes discovery tools based on neural networks,
clustering, genetic algorithms, and association rules
The problem
•
•
•
•
Large retailer
Over $1 billion in sales
Overstocked on slow-moving products
Under-stocked on most popular items at
critical selling periods.
Solution
• With Clustering and and NN:
– Review point-of-sale history and equate store
groupings to sales patterns.
– Forecast stocking requirements on a store-bystore basis.
Results
• Management is able to forecast seasonal
trends at the store-item level.
• The Decision Series tools showed that
clustering similar items into actionable groups
streamlined the ordering process.
• Revenues increased by 11.6%
Case #2
CHAID
Applied Metrix
• Uses a combination of CHAID segmentation
and logistic regression response probability
modeling to establish predictive models that
are deployed over a proprietary Internet
system
The problem
• Home equity marketer that extended home
equity lines of credit at the national level.
• The client’s goal was to increase the
efficiency of targeting current mortgage
customers who might be interested in the
client’s service.
The Solution
• CHAID identified 16
distinct market
segments.
• In particular, one
particular segment
accounted for 65% of
responses to the
mailing.
Results
• The highest-rated group from the predictive model
had by far the highest response rate to the equity
line of credit campaign—85% above average for the
direct mailing,
• The goal of the program was a 10% increase in
response rate, but the actual response rate
increased 30%.
• The firm was able to increase profits by over one
million dollars in the first year after implementation.
CHAID vs. Neural Networks
Clarity and explicability
- CHAID model is understandable as a set of rules
- Neural Network is obscure
Implementation/integration
- The CHAID model is much easier to be implemented
that a Neural Network.
- The risk of missing code by an IT department is slim
for a CHAID model and higher for a Neural Network.
Data Requirements
- The data for both techniques requires some preprocessing.
- Neural Network require the data be transformed into
binary format.
Accuracy of model
- Neural Networks provide more accurate models,
especially for complex problems.
Construction of model
- CHAID is easier and quicker to construct.
- Neural Networks have many parameters that must
be set and require more skilled manipulation.
Cost
- Building a Neural Network is more costly then
building a CHAID model.
Aplications
- CHAID and Neural Networks can create predictive
models.
- Neural Networks can handle both categorical and
continuous independent variables, but these have to
be transformed to 0/1 input variables.
- When all or most of the independent variables are
continuous, neural networks should perform better
than CHAID.
Aplications
- The Neural Networks and CHAID can be used to
solve sequence prediction problems.
- Neural Networks can be used to solve estimation
problems.
- CHAID provides good solutions to classification
problems, can be used for exploratory analysis and
can provide descriptive rules.
- An interesting development is the combination of
these two techniques to create “neural trees”.
CONCLUSIONS
- The choice among different options
is not as the choice to use data
mining technologies in a CRM
initiative.
- Data Mining represents the link
from the data stored over many
years through various interactions
with customers in diverse
situations, and the knowledge
necessary to be successful in
relationship marketing concepts.
CONCLUSIONS
- Through the full implementation of a CRM program,
which must include data mining, organizations foster
improved loyalty, increase the value of their
customers, and attract the right customers.
- As customers and businesses interact more
frequently, businesses will have to leverage on CRM
and related technologies to capture and analyze
massive amounts of customer information.
CONCLUSIONS
- CRM solutions focus
primarily on analyzing
consumer information for
economic benefits, and
very little touches on
ensuring privacy.