0% found this document useful (0 votes)
57 views9 pages

Lecture 5. Data Mining For Business Intelligence

Clustering and profiling are unsupervised data mining techniques used for pattern discovery in business intelligence. Clustering groups similar data records into clusters based on input variables without pre-defined groups, while profiling names each cluster based on descriptive variables. For example, retail customer data may be clustered and profiled into segments like "bargain hunters" or "impulse shoppers" based on demographic and shopping behavior variables. The goal is to form natural groupings of similar records to aid in future marketing strategies and segmenting markets. SAS Enterprise Miner software can be used to perform these clustering and profiling techniques.

Uploaded by

Kaitlyn Sheroke
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
57 views9 pages

Lecture 5. Data Mining For Business Intelligence

Clustering and profiling are unsupervised data mining techniques used for pattern discovery in business intelligence. Clustering groups similar data records into clusters based on input variables without pre-defined groups, while profiling names each cluster based on descriptive variables. For example, retail customer data may be clustered and profiled into segments like "bargain hunters" or "impulse shoppers" based on demographic and shopping behavior variables. The goal is to form natural groupings of similar records to aid in future marketing strategies and segmenting markets. SAS Enterprise Miner software can be used to perform these clustering and profiling techniques.

Uploaded by

Kaitlyn Sheroke
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 9

Business Intelligence

Lecture 5. Data Mining for Business Intelligence


Clustering & Profiling

Meiling Jiang, PhD


[email protected]
Business Intelligence

Decision optimization What is the best decision?


Advanced
Predictive modeling What will happen next?
analytics
Competitive Advantage

Forecasting What if these trends continue? SAS–Enterprise Miner

Basic statistical analysis Why is this happening?


Basic
Reporting with early warning What actions are needed?
analytics

Dynamic reporting Where exactly are the problems?

Ad-hoc reporting How many, how often, where? Reporting

Basic reporting What happened?


e.g., SAS-VA

IS at Hankamer, Baylor 2
Data Mining for Business Intelligence

IS at Hankamer, Baylor 3
Data Mining: Pattern Discovery

 Two data mining techniques for “unsupervised” pattern discovery


 Clustering & profiling

 Association (market basket) & sequence analysis

 Unsupervised means…
 No target variable (no dependent variable) with which to associate other variables
 Not a predictive method
 No pre-defined classes or groups

IS at Hankamer, Baylor 4
Clustering/Segmenting

 A pattern recognition method


 Unsupervised learning
• Not predictions
• No pre-defined groups

 Purpose
• To group data records based on similarities of their input (independent) variables

 Results
• Groupings are Clusters or Segments.

Input Input Input Input Grouping


Cluster 1

Cluster 3

Data records Cluster 2

Cluster 2

Cluster 1

IS at Hankamer, Baylor 5
An Example: Segmentation of Retail Customers

 Identifying customer segments


 While you have thousands of customers, there are really only a few major types into
which most customers can be grouped

 Bargain hunter
 Man/woman on a mission
 Impulse shopper
 Weary parent
 DINK: dual income, no kids

 What customer variables would be useful to make these customer


groups?

IS at Hankamer, Baylor 6
An Example: Segmentation of Retail Customers

 Analysis 1: clustering
 Customer data records are grouped (clustered) by common characteristics of input
variables

 Analysis 2: profiling
 Each cluster is named (profiled) according to input variables best describing that
cluster

 Expected results
 Cluster 1: bargain hunter
 Cluster 2: man/woman on a mission
 Cluster 3: impulse shopper
 Cluster 4: weary parent
 Cluster 5: DINK

IS at Hankamer, Baylor 7
Clustering & Profiling: The Main Idea

 Goal: Form groups (clusters) of similar records


 Analysis 1. clustering: group data records into distinct segments based on
demographic variables and/or behaviors
 Analysis 2. profiling: create profiles or descriptive tags based on the variables used to
cluster the data records
 Used for future marketing strategies
 Useful in segmenting markets

IS at Hankamer, Baylor 8
SAS Enterprise Miner

IS at Hankamer, Baylor 9

You might also like