Data Mining

1
Data Mining

• Getting to Know You!

2
Data Mining

• Data mining allows for the discovery of hidden
  patterns and relationships in large amounts of
  data.
• Data mining uses powerful analytic technologies
  to quickly and thoroughly explore mountains of
  data, isolating the valuable, usable information
  the business intelligence
• Ex. Data mining tells you which prospects are
  likely to become profitable customers and which
  are most likely to respond to your offer. ROI is
  increased by making offers to only those
  prospects likely to respond and become valuable
  customers.
• Spyware is any technology that aids in gathering
  information about a person or organization
  without their knowledge. On the Internet (where
  it is sometimes called a spybot or tracking
  software), spyware is programming that is put in
  someone's computer to secretly gather information
  about the user and relay it to advertisers or
  other interested parties. Often done via adware
  applications. Free spyware scan
• SPSS edu

3
What Business Problems does Data Mining Solve?

• You can use data mining to solve almost any
  business problem that involves data, including
• Increasing business unit and overall
  profitability
• Understanding customer desires and needs
• Identifying profitable customers and acquiring
  new ones
• Retaining customers and increasing loyalty
• Increasing ROI and reducing costs on promotions
• Cross-selling and up-selling
• Detecting fraud, waste and abuse
• Determining credit risks
• Increasing Web site profitability
• Increasing store traffic and optimizing layouts
  for increased sales
• Monitoring business performance SPSS

4
Data Mining Definitions and Uses

• Data mining refers to a wide range of techniques
  that look at underlying patterns or associations
  among elements within large data sets. These
  patterns are then used to form rules or
  guidelines for use in a wide range of marketing
  decisions. Ex. Insightful Miner demo
• Data mining tools can improve marketing
  management decisions such as
• inventory planning/management
• space utilization
• promotion management
• segmentation and target marketing
• improving sales force performance
• customer relationship management (CRM)
• and many others

5
CRISP-DM
Figure Phases of the CRISP-DM Process
Model Cross Industry Standard Process for Data
Mining Project Overview
6
What Can be Done with Data Mining?

• Through various algorithms, data mining software
  sorts through thousand of data points, organizes
  it, then summarizes complex relationships for the
  user.
• Data mining software typically follows one of
  five different analytical approaches
• Reporting and OLAP
• Associations
• Classifications
• Sequential patterns
• Clustering

7
Reporting and Online Analytical Processing (OLAP)

• Reporting (a.k.a. summary methods or baby stats)
  is one of the most basic, but extremely useful,
  techniques for data analysis.
• Provides simple views of the data such as counts,
  sums, percentages, and averages.
• Sample query How many units did we sell last
  month?
• OLAP (think multi-dimensional cross-tabulation)
  is useful because it provides cubes of
  reports that can break down one variable by
  another.
• Differs from traditional cross-tabs because it is
  interactive and you can drill down through the
  live reports to get more specific views of each
  cube (cell).
• See SPSS example in class.

8
Traditional Cross-tab vs. OLAP
Days per week SUBHT Cross-tabulation
Count
SUBHT
Days per week
no
yes
Total
daily
56
114
170
2-3 times
90
116
206
once
56
42
98
Sunday
41
132
173
5
1
1
2
Total
244
405
649
9
Associations

• The basic premise of associations is to find all
  relationships such that the presence of one set
  of items in a transaction implies other items,
  while controlling for extraneous factors
• Ex. 75 of consumers who buy beer also buy corn
  chips 26 of consumers who buy beer and corn
  chips also buy salsa
• Easy to do simple data analyses to discover these
  relationships
• Data mining software can simultaneously control
  for other variables that impact these
  relationships. Ex. Sale items, competitor
  actions, etc. to get true effects.

10
Classification

• Classification or profile generation uses data to
  develop profiles of different groups.
• Can be used for segmenting and targeting, market
  evaluation, product management, etc.
• Typically uses historical data to form rules that
  define groups. Those rules are then applied to
  new data to find similar groups.
• Ex. Based on past results, a hot prospect is a
  person who has an advanced degree, earns 150K or
  more, has made three online purchases over the
  last month, and has purchased computer related
  equipment within the past year. Find person who
  fit that profile and you have a good prospect.

11
Sequential Patterns

• This technique looks at purchases, or events,
  occurring in a sequence over time and tries to
  uncover patterns of behavior.
• Greatly useful for tracking effects of
  promotional activities also for work force
  allocation, inventory management, and
  pricing/valuation.
• Ex., Through data mining, company notices that
  when a customer buys a new DVD player, 85 of
  them return within three months to purchase
  speakers.
• Ex., a company with a brand new potato chip is
  trying to decide what time of the year to launch
  it. Might look for events, trends, etc.

12
Clustering

• Clustering will segment a database into subsets
  or clusters, creating a set of groups which have
  the maximum similarity within them and the
  maximum difference between them.
• Allows for consideration of multiple variables
  simultaneously.
• Great for building consumer segments, perceptual
  mapping, brand image assessment, etc.
• Ex. A company gathers consumer opinions on 25
  product attributes, then develops clusters based
  on those attributes
• Web Miner - data mining consultancy utilizing
  cluster analysis with pre-mined demographic
  information

13
Data Mining Tools/Web Mining

• Web Groove -solutions turn visitors into valued
  customers using adaptive personalization
  technology and click-stream analysis to increase
  ROI how it works
• SPSS Statistical Software utilizes
  cross-tabulation, OLAP cube, Crystal Reports etc.
  
• Microsoft Excel Spreadsheet
• Oracle data warehouse
• Quadstone uses a combination of analytical
  models including cross tabulation, OLAP, decision
  trees, scorecards. - demo available
• Web Miner -cluster analysis utilizing pre-mined
  demographic information demo
• DataDistilleries Dutch data mining company
• CRISP-DM Cross Industry Standard Process for
  Data Mining Project

14
Data Mining Leads to Business Intelligence

• Using Web-enabled business intelligence
  technology and applications, businesses are
  learning more about their best customers, their
  supply chains and product life cycles - and fast!
  
• Business intelligence delivers targeted,
  results-driven decisions and execution that can
  lead to competitive advantage.
• Business intelligence and data warehousing
  infrastructure can empower employees!