Applying Predictive Analytics: Finding Value in Data

ISBN-13: 9783030830724 / Angielski / Miękka / 2022 / 274 str.

Richard V. McCarthy; Mary M. McCarthy; Wendy Ceccucci

Opis

The new edition of this textbook presents a practical, updated approach to predictive analytics for classroom learning. The authors focus on using analytics to solve business problems and compares several different modeling techniques, all explained from examples using the SAS Enterprise Miner software. The authors demystify complex algorithms to show how they can be utilized and explained within the context of enhancing business opportunities. Each chapter includes an opening vignette that provides real-life examples of how business analytics have been used in various aspects of organizations to solve issues or improve their results. A running case provides an example of a how to build and analyze a complex analytics model and utilize it to predict future outcomes. The new edition includes chapters on clusters and associations and text mining to support predictive models. An additional case is also included that can be used with each chapter or as a semester project.

Spis treści

Chapter 1

Introduction to Predictive Analytics1

1.1 Predictive Analytics in Action2

1.2 Analytics Landscape8

1.3 Analytics

1.3.2 Predictive Analytics

1.4 Regression Analysis

1.5 Machine Learning Techniques

1.6 Predictive Analytics Model

1.7 Opportunities in Analytics

1.8 Introduction to the Automobile Insurance Claim Fraud Example

1.9 Chapter Summary

References

Chapter 239

Know Your Data – Data Preparation39

2.1 Classification of Data40

2.1.1 Qualitative versus Quantitative

2.1.2 Scales of Measurement

2.2. Data Preparation Methods.

2.2.1 Inconsistent Formats

2.2.2 Missing Data

2.2.3 Outliers

2.2.4 Other Data Cleansing Considerations

2.3 Data Sets and Data Partitioning

2.4 SAS Enterprise Miner™ Model Components

2.4.1 Step 1. Create Three of the Model Components

2.4.2 Step 2. Import an Excel File and Save as a SAS File

2.4.3 Step 3.  Create the Data Source

2.4.4 Step 4. Partition the Data Source

2.4.5 Step 5 Data Exploration

2.4.6 Step 6 Missing Data

2.4.7 Step 7. Handling Outliers

2.4.8 Step 8. Categorical Variables with Too Many Levels

2.5 Chapter Summary

References

Chapter 35

What do Descriptive Statistics Tell Us

3.1 Descriptive Analytics

3.2 The Role of the Mean, Median and Mode

3.3 Variance and Distribution

3.4 The Shape of the Distribution

3.4.2 Kurtosis

3.5 Covariance and Correlation

3.6 Variable Reduction

3.6.1 Variable Clustering

3.6.2 Principal Component Analysis

3.7 Hypothesis Testing2

3.8 Analysis of Variance (ANOVA)5

3.9 Chi Square6

3. Fit Statistics8

3. Stochastic Models9

3.12 Chapter Summary1

References2

Chapter 4

Predictive Models Using Regression5

4.1 Regression6

4.1.1 Classical assumptions7

4.2 Ordinary Least Squares8

4.3 Simple Linear Regression8

4.3.1 Determining Relationship Between Two Variables9

4.3.2 Line of Best Fit and Simple Linear Regression Equation9

4.4 Multiple Linear Regression1

4.4.1 Metrics to Evaluate the Strength of the Regression Line2

4.3.2 Best-fit model3

4.3.3 Selection of Variables in Regression3

4.5 Principal Component Regression5

4.5.1 Principal Component Analysis Revisited5

4.5.2 Principal Component Regression6

4.6 Partial Least Squares6

4.7 Logistic Regression7

4.7.1 Binary Logistic Regression8

4.7.2 Examination of Coefficients1

4.7.3 Multinomial Logistic Regression3

4.7.4 Ordinal Logistic Regression3

4.8 Implementation of Regression in SAS Enterprise Miner™3

4.8.1 Regression Node Train Properties: Class Targets4

4.8.2 Regression Node Train Properties: Model Options5

4.8.3 Regression Node Train Properties: Model Selection6

4.9 Implementation of Two-Factor Interaction and Polynomial Terms8

4.9.1 Regression Node Train Properties: Equation8

4. DMINE Regression in SAS Enterprise Miner™0

4..1 DMINE Properties0

4..2 DMINE Results2

4. Partial Least Squares Regression in SAS Enterprise Miner™4

4..1 Partial Least Squares Properties4

4..2 Partial Least Squares Results7

4. Least Angles Regression in SAS Enterprise Miner™9

4..1 Least Angle Regression Properties0

4..2 Least Angles Regression Results1

4. Other Forms of Regression4

4. Chapter Summary6

References9

Chapter 5

The Second of the Big Three – Decision Trees1

5.1 What is a Decision Tree?2

5.2 Creating a Decision Tree4

5.3 Data Partitions and Decision Trees6

5.4 Creating a Decision Tree Using SAS Enterprise Miner™9

The key properties include:5

Subtree Properties5

5.4.1 Overfitting1

5.5 Creating an Interactive Decision Tree using SAS Enterprise Miner ™1

5.6 Creating a Maximal Decision Tree using SAS Enterprise Miner ™6

5.7 Chapter Summary9

References1

Chapter 6

The Third of the Big Three - Neural Networks3

6.1 What is a Neural Network?4

6.2 History of Neural Networks6

6.3 Components of a Neural Network8

6.4 Neural Network Architectures2

6.5 Training a Neural Network5

6.6 Radial Basis Function Neural Networks6

6.7 Creating a Neural Network using SAS Enterprise MinerÔ7

6.8 Using SAS Enterprise MinerÔ to Automatically Generate a Neural Network0

6.9 Explaining a Neural Network6

6. Chapter Summary0

References3

Chapter 7

Model Comparisons and Scoring5

7.1 Beyond the Big

7.2 Gradient Boosting6

7.3 Ensemble Models0

7.4 Random Forests2

7.6 Two-Stage Model8

7.7 Comparing Predictive Models0

7.7.1 Evaluating Fit Statistics – Which Model Do We Use?2

7.8 Using Historical Data to Predict the Future – Scoring5

7.8.1 Analyzing and Reporting Results8

7.8.2 Save Data Node9

7.8.3 Reporter Node0

7.9 The Importance of Predictive Analytics2

7.9.1 What Should We Expect for Predictive Analytics in the Future?3

7. Chapter Summary4

References6

Chapter 8

finding Associations in Data through Cluster Analysis9

8.1 Applications and Uses of Cluster Analysis9

8.2 Types of Clustering Techniques0

8.3 Hierarchical Clustering1

8.3.1 Agglomerative Clustering1

8.3.2 Divisive Clustering1

8.3.3 Agglomerative vs Divisive Clustering6

8.4 Non-hierarchical clustering7

8.4.1 K-means Clustering7

8.4.2 Initial Centroid Selection1

8.4.3 Determining the Number of Clusters2

8.4.4 Evaluating your clusters5

8.5 Hierarchical vs Nonhierarchical6

8.6 Cluster Analysis using SAS Enterprise Miner™6

8.6.1 Cluster Node7

8.6.2 Additional Key Properties of the Cluster Node8

8.7 Applying Cluster Analysis to the Insurance Claim Fraud Data Set9

8.8 Chapter Summary8

References9

9.1 What is Text Analytics?1

9.2 Information Retrieval2

9.3 Text Parsing5

9.4 Zipf’s Law8

9.5 Text Filter9

9.6 Text Cluster1

9.7 Text Topic4

9.8 Text Rule Builder7

9.9 Text Profile8

9. Chapter Summary9

Discussion Questions0

References1

Appendix A3

Data Dictionary for the Automobile Insurance Claim Fraud Data Example3

Appendix B5

Can you Predict the Money Laundering Cases?5

B.1 Introduction5

B.2. Business Problem8

B.3. Analyze Data9

B.4. Development and Optimization of a Best Fit Model2

B.5. Final Report3

References4

 

O autorze

Richard V. McCarthy (DBA, Nova Southeastern University, MBA, Western New England College) is a professor of Computer Information Systems at the School of Business, Quinnipiac University. Prior to this, Dr. McCarthy was an associate professor of management information systems at Central Connecticut State University. He has twenty years of experience within the insurance industry and has held a Charter Property Casualty Underwriter (CPCU) designation since 1991. He has authored numerous research articles and contributed to several textbooks. He has served as the associate dean of the School of Business, the MBA director, and the director of the Master of Science in Business Analytics program. In 2019, he was awarded the Computer Educator of the Year from the International Association for Computer Information Systems. 

Wendy Ceccucci (PhD and MBA, Virginia Polytechnic University) is a Professor and Chair of Computer Information Systems at Quinnipiac University.  Her teaching areas include business analytics and programming. She is the past president of the Education Special Interest Group (EDSIG) of the Association for Information Technology Professionals (AITP) and past Associate Editor of the Information Systems Education Journal (ISEDJ). Her research interests include Information Systems Pedagogy. 

Mary McCarthy (DBA, Nova Southeastern University, MBA, University of Connecticut) is a professor and chair of Accounting, Central Connecticut State University. She has twenty years of financial reporting experience and has served as the controller for a Fortune 50 industry organization. She holds a CPA and CFA designation.  She has authored numerous research articles.