Meet the Authors

Sanjiv JaggiaProfessor of Economics

and FinanceCalifornia Polytechnic

State University

Alison KellyProfessor of Economics

Suffolk University

Kevin LertwacharaProfessor of Information

SystemsCalifornia Polytechnic

State University

Leida ChenProfessor of Information

SystemsCalifornia Polytechnic

State University

Innovative Methods for Teaching Business Analytics• Business analytics

• The scientific process of transforming data into insights for the purpose of making better decisions.

• Limitations of current business analytics pedagogy:• Heavy focus on the modeling phase only• Emphasis on technical analytical skillsets at the expense of storytelling • Students not adequately trained to deal with real life data and projects

• Infusion of the CRISP-DM framework in business analytics pedagogy (S. Jaggia, A. Kelly, K. Lertwachara, & L. Chen, 2020, “Applying the CRISP-DM Framework for Teaching Business Analytics,” Decision Sciences Journal of Innovative Education, 18(4), pp. 612 –634)

The CRISP-DM Framework

• CRISP-DM Learning Objectives:

• Business understanding

• Data understanding• Data wrangling• Modeling• Evaluation• Storytelling

Project Learning Objectives

CRISP-DM Phases Project Learning Objectives

Business UnderstandingFormulate business questions that lead to business strategies or actions.

Data Understanding Describe the data in terms of the business context.

Data Preparation Perform data wrangling to prepare the data for subsequent analyses.

ModelingDevelop predictive models to inform decision-making and select the best predictive model(s).

Evaluation Evaluate model performance from the business perspective.

Deployment Communicate key findings through storytelling.

The Data• ERIM data set provided by the University

of Chicago’s Booth School of Business (https://www.chicagobooth.edu/research/kilts/datasets/erim)

• 3,189 households in two Midwestern cities and their purchases in a number of product categories (e.g. frozen dinners, yogurt, ketchup, margarine, etc.)

• Project scope:• One product category per team• Competition among teams on the same

product category or categories

Variable DescriptionHH_ID The household’s identification number

ResType Types of residence: 1 for Apartment, 2 for Condo, 3 for Single Family, 4 for Multiple Family, 5 for Mobile, and 6 for Other.

ResStatus Residence status: 1 for owned home, 2 for rented, and 3 for other.

HHInc The average annual income of a household; there are 14 categories for this variable.

HHNbr The number of members in the household.

.

.

.

.

.

.

MWrkHrs The average hours worked each week by the male head of household.

MBirth The birth year of the male head of household.

M_RelRelationship within the household: 1 for female head of household, 2 for male head of household, 3 for daughter, 4 for son, and 5 for other.

Dogs Whether or not the household has dogs; 1 if yes, 0 otherwise.YogExp A household’s yogurt expenditures (in $)DinExp A household’s frozen dinner expenditures (in $)

Modeling and AssessmentVariable Yogurt Frozen Dinners

Intercept 0.3047(0.447)

‒0.0870(0.814)

HH Income 0.0582 **(0.007)

‒0.0387 **(0.025)

Cable ‒0.0052(0.959)

‒0.0784(0.349)

Single Family Home ‒0.0614(0.714)

0.1367(0.365)

Own Home 0.3565 **(0.022)

0.1359(0.326)

Pets 0.0740 (0.456)

0.2186 **(0.007)

Married 0.6901 **(0.004)

0.4194(0.109)

College Educated Both 0.6808 **(0.001)

‒0.4930 **(0.000)

College Educated One 0.3977 *(0.073)

‒0.2248(0.250)

Work Hours 0.0014(0.717)

‒0.0045(0.165)

Other HH Members 0.2455 **(0.000)

0.1735 **(0.000)

Age ‒0.0095 **(0.044)

‒0.0214 **(0.000)

Female HH 0.9031**(0.000)

0.1424(0.585)

ME RMSE MAE

-0.0048 1.9807 1.6155

MeasureYogurt Frozen Dinners

Cutoff = 0.5 Cutoff = 0.82 Cutoff = 0.5 Cutoff = 0.33Accuracy 0.8141 0.5890 0.6541 0.5890Sensitivity 0.9971 0.5833 0.1106 0.5671Specificity 0.0085 0.6144 0.9259 0.6000

• Evaluate model performance using the validation or test data sets

• Predictive performance measures vs. goodness-of-fit statistics

• Business Understanding (Chapter 1 plus Intro Case and Big Data Cases in all chapters)

• Data Preparation and Understanding (Data Wrangling, Visualization, and Summary Measures; Chapters 2, 3)

• Modeling (Predictive (Chapters 6-12) and Prescriptive (Chapter 13))

• Evaluation (Intro Case and Big Data Cases; all chapters)

• Storytelling (Case Synopsis and Writing with Big Data; all chapters)

A Textbook that Applies CRISP-DM

Emphasis on Communication

• Integrated Introductory Case and SynopsisEach chapter opens with a real-life case study that forms the basis for several

examples within the chapter. A synopsis is presented once the questions pertaining to the case have been

answered.

• Writing with Big DataThroughout the text, we use big data sets to help introduce problems,

formulate possible solutions, and communicate the findings, based on the concepts introduced in the chapters.

Detailed Software Instructions

Dedicated Chapter on Data Wrangling

Topics include:

• Key concepts related to data management

• Data inspection

• Binning, subsetting, and treatment of missing values and outliers

• Transformation of numeric variables

• Transformation of categorical variables

Writing with Big DataSales_Rep Business Age Female … Personality Certificates … Salary NPS 1 Hardware 59 1 … Diplomat 1 … 70200 5 2 Hardware 52 0 … Diplomat 4 … 133000 10 … … … 21990 Software 23 1 … Explorer 1 … 47400 5

Emphasis on Data Mining

• In addition to two chapters on linear and logistic regression and a chapter on forecasting, there are four exclusive chapters on data mining, includingIntroduction to data mining: distance measures, performance

evaluation, and principal component analysis (PCA)Supervised learning: k-nearest neighbors (KNN) and naïve Bayes,

classification and regression trees, and ensemble treesUnsupervised learning: hierarchical and k-means clustering, and

association rules

• Over 200 exercises in these four exclusive chapters.

McGraw Hill’s Connect

• Fully engaged with the development of the Connect product

• Separate questions for Analytic Solver and R, when needed

• Careful with the choice of:Tolerance limits

Algorithmic exercises

Different ways to use the Text

• One TermBusiness Statistics with Analytics Flavor (Chapters 1, 2, 3, 4, 5, 6, 7, 12)

Holistic Approach to Business Analytics (Chapters 1, 2, 3, 6, 7, 8, 9, 10, and 11)

• Two Terms Term 1: Chapters 1, 2, 3, 4, 5, 6, 12

Term 2: Chapters 7, 8, 9, 10, 11, 13

• Applicable for Undergraduate and Graduate Courses

Using the Text for Online Instruction

Example: undergraduate business analytics course• Coverage: Chapters 1, 2, 3, 6, 7, 8, 9, 10, and 11

• Asynchronous resources: SmartBook, ReadAnywhere mobile app, Connect

• With detailed software instructions in the text, synchronous meetings can focus more on coaching, consultation, and interpretation of results

• Instructor-defined options (e.g., time limit, attempts, score reduction, access to hints/eBook and solutions )

• Performance reports by assignment and by student on Connect

Second Edition• Expanded coverage on prescriptive analytics

• Spreadsheet Modeling; Simulation and Risk Analysis; Optimization: Linear Programming; Optimization: Integer and Nonlinear Programming

• More in-depth coverage on selected topics:• Logistic Regression; Data Visualization (including an appendix on Tableau)

• Chapter sequence rearranged based on feedback to allow more flexibility in course design

• Simplification and affordability of software tools• Most chapters can be taught using Excel and R• Any version of R, including any future versions, would work

• Additional and updated exercises and cases including those related to the current events.

Questions, Comments, and Suggestions?