Business Intelligence:

A Managerial Perspective on
Analytics (3rd Edition)

Chapter 4:
Data Mining
Learning Objectives
 Define data mining as an enabling technology for
business intelligence
 Understand the objectives and benefits of business
analytics and data mining
 Recognize the wide range of applications of data
mining
 Learn the standardized data mining processes
 CRISP-DM
 SEMMA
 KDD

(Continued…)
Copyright © 2014 Pearson Education, Inc. Slide 4-
1- 2
Learning Objectives
 Understand the steps involved in data
preprocessing for data mining
 Learn different methods and algorithms of data
mining
 Build awareness of the existing data mining
software tools
 Commercial versus free/open source
 Understand the pitfalls and myths of data mining

Copyright © 2014 Pearson Education, Inc. Slide 4-

1- 3
Opening Vignette…
Cabela’s Reels in More Customers
with Advanced Analytics and Data Mining
 Decision situation
 Problem
 Proposed solution
 Results
 Answer & discuss the case questions.

Copyright © 2014 Pearson Education, Inc. Slide 4- 4

Questions for the Opening Vignette
1. Why should retailers, especially omni-channel retailers,
pay extra attention to advanced analytics and data mining?
2. What are the top challenges for multi-channel retailers?
Can you think of other industry segments that face similar
problems/challenges?
3. What are the sources of data that retailers such as
Cabela’s use for their data mining projects?
4. What does it mean to have a “single view of the
customer”? How can it be accomplished?
5. What type of analytics help did Cabela’s get from their
efforts? Can you think of any other potential benefits of
analytics for large-scale retailers like Cabela’s?
Copyright © 2014 Pearson Education, Inc. Slide 4-
1- 5
Data Mining Concepts and
Definitions
Why Data Mining?
 More intense competition at the global scale.
 Recognition of the value in data sources.
 Availability of quality data on customers, vendors,
transactions, Web, etc.
 Consolidation and integration of data repositories into data
warehouses.
 The exponential increase in data processing and storage
capabilities; and decrease in cost.
 Movement toward conversion of information resources into
nonphysical form.

Copyright © 2014 Pearson Education, Inc. Slide 4- 6

Definition of Data Mining
 The nontrivial process of identifying valid, novel,
potentially useful, and ultimately understandable
patterns in data stored in structured databases. -
Fayyad et al., (1996)
 Keywords in this definition: Process, nontrivial, valid,
novel, potentially useful, understandable.
 Data mining: a misnomer?
 Other names: knowledge extraction, pattern analysis,
knowledge discovery, information harvesting, pattern
searching, data dredging,…

Copyright © 2014 Pearson Education, Inc. Slide 4- 7

Data Mining at the Intersection of
Many Disciplines

Pattern
Recognition

DATA Machine
MINING Learning

Mathematical
Modeling Databases

Management Science &

Information Systems

Copyright © 2014 Pearson Education, Inc. Slide 4- 8

Data Mining
Characteristics/Objectives
 Source of data for DM is often a consolidated data
warehouse (not always!).
 DM environment is usually a client-server or a
Web-based information systems architecture.
 Data is the most critical ingredient for DM which
may include soft/unstructured data.
 The miner is often an end user
 Striking it rich requires creative thinking
 Data mining tools’ capabilities and ease of use are
essential (Web, Parallel processing, etc.)
Copyright © 2014 Pearson Education, Inc. Slide 4- 9
Data in Data Mining
 Data: a collection of facts usually obtained as the result of
experiences, observations, or experiments.
 Data may consist of numbers, words, images, …
 Data: lowest level of abstraction (from which information and
knowledge are derived).
Data
- DM with
different data
types?
Categorical Numerical
- Other data
types?
Nominal Ordinal Interval Ratio

Copyright © 2014 Pearson Education, Inc. Slide 4- 10

What Does DM Do?
How Does it Work?
 DM extract patterns from data
 Pattern? A mathematical (numeric and/or
symbolic) relationship among data items

 Types of patterns
 Association
 Prediction
 Cluster (segmentation)
 Sequential (or time series) relationships
Copyright © 2014 Pearson Education, Inc. Slide 4- 11
A Taxonomy for Data Mining Tasks
Data Mining Learning Method Popular Algorithms

Classification and Regression Trees,

Prediction Supervised
ANN, SVM, Genetic Algorithms

Decision trees, ANN/MLP, SVM, Rough

Classification Supervised
sets, Genetic Algorithms

Linear/Nonlinear Regression, Regression

Regression Supervised
trees, ANN/MLP, SVM

Association Unsupervised Apriory, OneR, ZeroR, Eclat

Link analysis Unsupervised Expectation Maximization, Apriory

Algorithm, Graph-based Matching

Sequence analysis Unsupervised Apriory Algorithm, FP-Growth technique

Clustering Unsupervised K-means, ANN/SOM

Outlier analysis Unsupervised K-means, Expectation Maximization (EM)

Copyright © 2014 Pearson Education, Inc. Slide 4- 12

Data Mining Tasks (cont.)
 Time-series forecasting
 Part of sequence or link analysis?
 Visualization
 Another data mining task?

 Types of DM
 Hypothesis-driven data mining
 Discovery-driven data mining

Copyright © 2014 Pearson Education, Inc. Slide 4- 13

Data Mining Applications
 Customer Relationship Management
 Maximize return on marketing campaigns
 Improve customer retention (churn analysis)
 Maximize customer value (cross-, up-selling)
 Identify and treat most valued customers

 Banking & Other Financial

 Automate the loan application process
 Detecting fraudulent transactions
 Maximize customer value (cross-, up-selling)
 Optimizing cash reserves with forecasting

Copyright © 2014 Pearson Education, Inc. Slide 4- 14

Data Mining Applications (cont.)
 Retailing and Logistics
 Optimize inventory levels at different locations
 Improve the store layout and sales promotions
 Optimize logistics by predicting seasonal effects
 Minimize losses due to limited shelf life

 Manufacturing and Maintenance

 Predict/prevent machinery failures
 Identify anomalies in production systems to optimize
the use manufacturing capacity
 Discover novel patterns to improve product quality

Copyright © 2014 Pearson Education, Inc. Slide 4- 15

Data Mining Applications (cont.)
 Brokerage and Securities Trading
 Predict changes on certain bond prices
 Forecast the direction of stock fluctuations
 Assess the effect of events on market movements
 Identify and prevent fraudulent activities in trading

 Insurance
 Forecast claim costs for better business planning
 Determine optimal rate plans
 Optimize marketing to specific customers
 Identify and prevent fraudulent claim activities

Copyright © 2014 Pearson Education, Inc. Slide 4- 16

Data Mining Applications (cont.)
 Computer hardware and software
 Science and engineering
 Government and defense
 Homeland security and law enforcement
 Travel industry
 Healthcare Highly popular
 Medicine application areas for
data mining
 Entertainment industry
 Sports
 Etc.

Copyright © 2014 Pearson Education, Inc. Slide 4- 17

Data Mining Process
 A manifestation of best practices
 A systematic way to conduct DM projects
 Different groups has different versions
 Most common standard processes:
 CRISP-DM (Cross-Industry Standard Process
for Data Mining)
 SEMMA (Sample, Explore, Modify, Model, and
Assess)
 KDD (Knowledge Discovery in Databases)
Copyright © 2014 Pearson Education, Inc. Slide 4- 18
Data Mining Process

Source: KDNuggets.com
Copyright © 2014 Pearson Education, Inc. Slide 4- 19
Data Mining Process: CRISP-DM

1 2
Business Data
Understanding Understanding

3
Data
Preparation
Data Sources
6
4
Deployment
Model
Building

5
Testing and
Evaluation

Copyright © 2014 Pearson Education, Inc. Slide 4- 20

Data Mining Process: CRISP-DM
Step 1: Business Understanding Accounts for
~85% of total
Step 2: Data Understanding project time

Step 3: Data Preparation (!)

Step 4: Model Building
Step 5: Testing and Evaluation
Step 6: Deployment
 The process is highly repetitive and
experimental (DM: art versus science?)
Copyright © 2014 Pearson Education, Inc. Slide 4- 21
Data Preparation – A Critical DM Task
Real-world
Data

· Collect data
Data Consolidation · Select data
· Integrate data

· Impute missing values

Data Cleaning · Reduce noise in data
· Eliminate inconsistencies

· Normalize data
Data Transformation · Discretize/aggregate data
· Construct new attributes

· Reduce number of variables

Data Reduction · Reduce number of cases
· Balance skewed data

Well-formed
Data

Copyright © 2014 Pearson Education, Inc. Slide 4- 22

Data Mining Process: SEMMA
Sample
(Generate a representative
sample of the data)

Assess Explore
(Evaluate the accuracy and (Visualization and basic
usefulness of the models) description of the data)

SEMMA

Model Modify
(Use variety of statistical and (Select variables, transform
machine learning models ) variable representations)

Copyright © 2014 Pearson Education, Inc. Slide 4- 23

Data Mining Methods: Classification
 Most frequently used DM method
 Part of the machine-learning family
 Employ supervised learning
 Learn from past data, classify new data
 The output variable is categorical
(nominal or ordinal) in nature
 Classification versus regression?
 Classification versus clustering?
Copyright © 2014 Pearson Education, Inc. Slide 4- 24
Assessment Methods for Classification
 Predictive accuracy
 Hit rate
 Speed
 Model building; predicting
 Robustness
 Scalability
 Interpretability
 Transparency, explainability
Copyright © 2014 Pearson Education, Inc. Slide 4- 25
Accuracy of Classification Models
 In classification problems, the primary source for
accuracy estimation is the confusion matrix
True Class TP  TN
Positive Negative Accuracy 
TP  TN  FP  FN
True False TP
Positive

Positive Positive True Positive Rate 

Predicted Class

TP  FN
Count (TP) Count (FP)
TN
True Negative Rate 
TN  FP
Negative

False True
Negative Negative
Count (FN) Count (TN) TP TP
P recision  Recall 
TP  FP TP  FN

Copyright © 2014 Pearson Education, Inc. Slide 4- 26

Estimation Methodologies for
Classification
 Simple split (or holdout or test sample estimation)
 Split the data into 2 mutually exclusive sets training
(~70%) and testing (30%)

Model
Training Data Development
2/3

Preprocessed Classifier
Data
1/3 Model
Prediction
Assessment
Testing Data Accuracy
(scoring)

 For ANN, the data is split into three sub-sets (training

[~60%], validation [~20%], testing [~20%])

Copyright © 2014 Pearson Education, Inc. Slide 4- 27

Estimation Methodologies for
Classification
 k-Fold Cross Validation (rotation estimation)
 Split the data into k mutually exclusive subsets
 Use each subset as testing while using the rest of the
subsets as training
 Repeat the experimentation for k times
 Aggregate the test results for true estimation of prediction
accuracy training
 Other estimation methodologies
 Leave-one-out, bootstrapping, jackknifing
 Area under the ROC curve

Copyright © 2014 Pearson Education, Inc. Slide 4- 28

Estimation Methodologies for
Classification – ROC Curve
1

0.9

0.8
A
True Positive Rate (Sensitivity)

0.7

B
0.6

C
0.5

0.4

0.3

0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

False Positive Rate (1 - Specificity)

Copyright © 2014 Pearson Education, Inc. Slide 4- 29

Classification Techniques
 Decision tree analysis
 Statistical analysis
 Neural networks
 Support vector machines
 Case-based reasoning
 Bayesian classifiers
 Genetic algorithms
 Rough sets
Copyright © 2014 Pearson Education, Inc. Slide 4- 30
 Decision Trees
 Employs the divide and conquer method
 Recursively divides a training set until each
division consists of examples from one class
1. Create a root node and assign all of the training
A
data to it.
general
algorith 2. Select the best splitting attribute.
m for 3. Add a branch to the root node for each value of
decision the split. Split the data into mutually exclusive
tree subsets along the lines of the specific split.
building 4. Repeat the steps 2 and 3 for each and every leaf
node until the stopping criteria is reached.
Copyright © 2014 Pearson Education, Inc. Slide 4- 31
Decision Trees
 DT algorithms mainly differ on
1. Splitting criteria
 Which variable, what value, etc.
2. Stopping criteria
 When to stop building the tree
3. Pruning (generalization method)
 Pre-pruning versus post-pruning
 Most popular DT algorithms include
 ID3, C4.5, C5; CART; CHAID; M5

Copyright © 2014 Pearson Education, Inc. Slide 4- 32

Decision Trees
 Alternative splitting criteria
 Gini index determines the purity of a specific
class as a result of a decision to branch
along a particular attribute/value
 Used in CART
 Information gain uses entropy to measure
the extent of uncertainty or randomness of a
particular attribute/value split
 Used in ID3, C4.5, C5
 Chi-square statistics (used in CHAID)
Copyright © 2014 Pearson Education, Inc. Slide 4- 33
Cluster Analysis for Data Mining
 Used for automatic identification of
natural groupings of things
 Part of the machine-learning family
 Employ unsupervised learning
 Learns the clusters of things from past
data, then assigns new instances
 There is not an output variable
 Also known as segmentation
Copyright © 2014 Pearson Education, Inc. Slide 4- 34
Cluster Analysis for Data Mining
 Clustering results may be used to
 Identify natural groupings of customers
 Identify rules for assigning new cases to
classes for targeting/diagnostic purposes
 Provide characterization, definition, labeling
of populations
 Decrease the size and complexity of
problems for other data mining methods
 Identify outliers in a specific domain (e.g.,
rare-event detection)
Copyright © 2014 Pearson Education, Inc. Slide 4- 35
Cluster Analysis for Data Mining
 Analysis methods
 Statistical methods (including both
hierarchical and nonhierarchical), such as
k-means, k-modes, and so on.
 Neural networks (adaptive resonance
theory [ART], self-organizing map [SOM])
 Fuzzy logic (e.g., fuzzy c-means algorithm)
 Genetic algorithms

Copyright © 2014 Pearson Education, Inc. Slide 4- 36

Cluster Analysis for Data Mining
 How many clusters?
 There is not a “truly optimal” way to calculate it
 Heuristics are often used
 Most cluster analysis methods involve the
use of a distance measure to calculate the
closeness between pairs of items.
 Euclidian versus Manhattan/Rectilinear
distance

Copyright © 2014 Pearson Education, Inc. Slide 4- 37

Cluster Analysis for Data Mining
 k-Means Clustering Algorithm
 k : pre-determined number of clusters
 Algorithm (Step 0: determine value of k)
Step 1: Randomly generate k random points as initial
cluster centers.
Step 2: Assign each point to the nearest cluster center.
Step 3: Re-compute the new cluster centers.
Repetition step: Repeat steps 3 and 4 until some
convergence criterion is met (usually that the
assignment of points to clusters becomes stable).

Copyright © 2014 Pearson Education, Inc. Slide 4- 38

 Cluster Analysis for Data Mining -
k-Means Clustering Algorithm
Step 1 Step 2 Step 3

Copyright © 2014 Pearson Education, Inc. Slide 4- 39

Association Rule Mining
 A very popular DM method in business
 Finds interesting relationships (affinities) between
variables (items or events)
 Part of machine learning family
 Employs unsupervised learning
 There is no output variable
 Also known as market basket analysis
 Often used as an example to describe DM to
ordinary people, such as the famous “relationship
between diapers and beers!”
Copyright © 2014 Pearson Education, Inc. Slide 4- 40
Association Rule Mining
 Input: the simple point-of-sale transaction data
 Output: Most frequent affinities among items
 Example: according to the transaction data…
“Customer who bought a lap-top computer and a
virus protection software, also bought extended
service plan 70 percent of the time."
 How do you use such a pattern/knowledge?
 Put the items next to each other
 Promote the items as a package
 Place items far apart from each other!

Copyright © 2014 Pearson Education, Inc. Slide 4- 41

Association Rule Mining
 A representative applications of association rule
mining include
 In business: cross-marketing, cross-selling, store design,
catalog design, e-commerce site design, optimization of
online advertising, product pricing, and sales/promotion
configuration
 In medicine: relationships between symptoms and
illnesses; diagnosis and patient characteristics and
treatments (to be used in medical DSS); and genes and
their functions (to be used in genomics projects)
 …

Copyright © 2014 Pearson Education, Inc. Slide 4- 42

Association Rule Mining
 Are all association rules interesting and useful?

A Generic Rule: X  Y [S%, C%]

X, Y: products and/or services
X: Left-hand-side (LHS)
Y: Right-hand-side (RHS)
S: Support: how often X and Y go together
C: Confidence: how often Y go together with the X
Example: {Laptop Computer, Antivirus Software} 
{Extended Service Plan} [30%, 70%]

Copyright © 2014 Pearson Education, Inc. Slide 4- 43

Association Rule Mining
 Algorithms are available for generating
association rules
 Apriori
 Eclat
 FP-Growth
 + Derivatives and hybrids of the three
 The algorithms help identify the
frequent item sets, which are, then
converted to association rules
Copyright © 2014 Pearson Education, Inc. Slide 4- 44
Association Rule Mining
 Apriori Algorithm
 Finds subsets that are common to at least a
minimum number of the itemsets
 Uses a bottom-up approach
 frequent subsets are extended one item at a time
(the size of frequent subsets increases from one-
item subsets to two-item subsets, then three-item
subsets, and so on), and
 groups of candidates at each level are tested
against the data for minimum support.
(see the figure)  --
Copyright © 2014 Pearson Education, Inc. Slide 4- 45
 Association Rule Mining
Apriori Algorithm

Raw Transaction Data One-item Itemsets Two-item Itemsets Three-item Itemsets

Transaction SKUs Itemset Itemset Itemset

Support Support Support
No (Item No) (SKUs) (SKUs) (SKUs)

1 1, 2, 3, 4 1 3 1, 2 3 1, 2, 4 3
1 2, 3, 4 2 6 1, 3 2 2, 3, 4 3
1 2, 3 3 4 1, 4 3
1 1, 2, 4 4 5 2, 3 4
1 1, 2, 3, 4 2, 4 5
1 2, 4 3, 4 3

Copyright © 2014 Pearson Education, Inc. Slide 4- 46

Artificial Neural Networks
for Data Mining
 Artificial neural networks (ANN or NN) is a brain
metaphor for information processing
 a.k.a. Neural Computing
 Very good at capturing highly complex non-
linear functions!
 Many uses – prediction (regression, classification),
clustering/segmentation
 Many application areas - finance, medicine, marketing,
manufacturing, service operations, information systems, …

Copyright © 2014 Pearson Education, Inc. Slide 4- 47

 Dendrites Biological NN
Synapse
Synapse

Axon

Axon

Dendrites Neuron
Neuron

Biological x1 Artificial NN

versus Inputs
w1 Y1

Artificial x2
w2 Processing
Outputs

Neural .
Element (PE)


n
f (S )
Y Y2
. S  X iW
Networks Weights
i
Transfer .
. .
i 1
Function .
Summation

wn Yn
Biological Artificial
xn
Neuron Node (or PE)
Dendrites Input
Axon Output
Synapse Weight
Slow Fast
Many (109) Few (102)
Elements/Concepts of ANN
 Processing element (PE)
 Information processing
 Network structure
 Feedforward vs. recurrent vs. multi-layer…
 Learning parameters
 Supervised/unsupervised, backpropagation,
learning rate, momentum
 ANN Software – NN shells, integrated modules
in comprehensive DM software, …
Copyright © 2014 Pearson Education, Inc. Slide 4- 49
 SPSS PASW Modeler (formerly Clementine)

Data Mining RapidMiner

SAS / SAS Enterprise Miner

Software
Microsoft Excel

Your own code

Weka (now Pentaho)

 Commercial KXEN

MATLAB
 IBM SPSS Modeler Other commercial tools

(formerly Clementine) KNIME

Microsoft SQL Server

 SAS - Enterprise Miner Other free tools

 IBM - Intelligent Miner Zementis

Oracle DM

 StatSoft – Statistica Data Statsoft Statistica

Salford CART, Mars, other

Miner Orange

 … many more Angoss

C4.5, C5.0, See5

 Free and/or Open Source Bayesia

Insightful Miner/S-Plus (now TIBCO)

 RapidMiner Megaputer

Viscovery
 Weka Clario Analytics
Total (w/ others) Alone
 R, … Miner3D

Thinkanalytics

0 20 40 60 80 100 120
Source: KDNuggets.com

Copyright © 2014 Pearson Education, Inc. Slide 4- 50

Opening Vignette…
“Data Mining Goes to Hollywood!”
 Decision situation
 Problem
 Proposed solution
 Results
 Answer & discuss the case questions.

Copyright © 2014 Pearson Education, Inc. Slide 4- 51

 Opining Vignette:
Data Mining Goes to Hollywood!
Class No. 1 2 3 4 5 6 7 8 9

Range <1 >1 > 10 > 20 > 40 > 65 > 100 > 150 > 200
(in $Millions) (Flop) < 10 < 20 < 40 < 65 < 100 < 150 < 200 (Blockbuster)

Number of
Independent Variable Possible Values
Values
Depende
nt MPAA Rating 5 G, PG, PG-13, R, NR
Variable Independe Competition 3 High, Medium, Low
nt Star value 3 High, Medium, Low
Variables
Sci-Fi, Historic Epic Drama,
Modern Drama, Politically
A Typical Genre 10 Related, Thriller, Horror,
Comedy, Cartoon, Action,
Classification Documentary

Problem Special effects 3 High, Medium, Low

Sequel 1 Yes, No
Number of screens 1 Positive integer

Copyright © 2014 Pearson Education, Inc. Slide 4- 52

 Opining Vignette:
Data Mining Goes to Hollywood!
Model
Development

The DM
process

Process
Map in Model

IBM Assessment
process

SPSS
Modele
r

Copyright © 2014 Pearson Education, Inc. Slide 4- 53

Opining Vignette:
Data Mining Goes to Hollywood!
Prediction Models

Individual Models Ensemble Models

Performance Random Boosted Fusion

Measure SVM ANN C&RT Forest Tree (Average)

Count (Bingo) 192 182 140 189 187 194

Count (1-Away) 104 120 126 121 104 120

Accuracy (% Bingo) 55.49% 52.60% 40.46% 54.62% 54.05% 56.07%

Accuracy (% 1-Away) 85.55% 87.28% 76.88% 89.60% 84.10% 90.75%

Standard deviation 0.93 0.87 1.05 0.76 0.84 0.63

* Training set: 1998 – 2005 movies; Test set: 2006 movies

Copyright © 2014 Pearson Education, Inc. Slide 4- 54

Data Mining Myths
 Data mining …
 provides instant solutions/predictions
 is not yet viable for business applications
 requires a separate, dedicated database
 can only be done by those with advanced
degrees
 is only for large firms that have lots of
customer data
 is another name for the good-old statistics
Copyright © 2014 Pearson Education, Inc. Slide 4- 55
Common Data Mining Blunders
1. Selecting the wrong problem for data mining
2. Ignoring what your sponsor thinks data mining is
and what it really can/cannot do
3. Not leaving insufficient time for data acquisition,
selection and preparation
4. Looking only at aggregated results and not at
individual records/predictions
5. Being sloppy about keeping track of the data
mining procedure and results
6. …more in the book
Copyright © 2014 Pearson Education, Inc. Slide 4- 56
End of the Chapter

 Questions, comments

Copyright © 2014 Pearson Education, Inc. Slide 4- 57

 All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any
means, electronic, mechanical, photocopying, recording, or otherwise,
without the prior written permission of the publisher. Printed in the
United States of America.

Copyright © 2014 Pearson Education, Inc.

Copyright © 2014 Pearson Education, Inc. Slide 4-

1- 58