MS Analytics Course

ITEC 621 Predictive Analytics

Last updated 6/12/2017
[Go to Class Schedule]

Professor: J. Alberto Espinosa, Ph.D.

[email protected]
http://auapps.american.edu/alberto/www/
Office: KSB 33
Office Hours: T-Th 4:00 - 5:30 PM
Term: Summer 2017 (7-week module)
Class Schedule: Tu & Th 5:30 - 8:40 PM
Room: KSB T-61
Textbook
Required:
 “An Introduction to Statistical Learning: with Applications in R” by James, Witten, Hastie and
Tibshirani, Springer, 1st. Edition, 2013. Please note that the authors of this book have a free PDF
version on their website:
http://www-bcf.usc.edu/~gareth/ISL/ISLR%20First%20Printing.pdf
 ISLR Textbook authors’ lectures and videos:
http://fs2.american.edu/alberto/www/analytics/ISLRLectures.html
Optional (Recommended for R):
 “R for Everyone: Advanced Analytics and Graphics” by Lander, J., Addison-Wesley Data & Analytics
Series, 1 edition, 2013
Analytics Resources: http://fs2.american.edu/alberto/www/analytics/AnalyticsResources.html

Course Overview
Analytics is the process of transforming data into insight for making better decisions (INFORMS).
There are three primary types of analytics: “Descriptive,” which examines historical data and identifies
and reports historical patterns and trends; “Predictive,” which predicts outcomes and future trends
from existing data to help discover new relationships; “Prescriptive,” which formulates and evaluates
new ways for a business to operate. This course focuses on the second type, Predictive Analytics,
which is of particular importance for business because it helps decision makers evaluate possible

Last updated: 5/6/2016

1
 outcomes (e.g., revenues, profits, market
share, probability of making a sale, probability
of losing a client, etc.) based on other
historical data predictors (e.g., marketing
expenditures, quality assurance investments,
sales force size, etc.). The process of analytics
involves specifying a question, problem, or
decision, and finding the right answers using
data. The process begins with identifying the
appropriate data sources (internal or external,
data format), and the appropriate models,
tools, and methods for analysis. In this
course, students are introduced to predictive modeling methods, approaches and tools. Students
develop skills in predictive analytics that will allow them to: (1) develop and use advanced predictive
analytics methods; (2) develop expertise in the use of popular tools and software for predictive
analytics; (3) learn how to develop predictive analytics questions, identify and select the most
appropriate predictive analytics methods and tools, apply these methods to answer the respective
questions and presenting data-driven solutions.
Course Learning Objectives
After completing this class, the student will develop the following competencies.
 Competency-1: Predictive Analytics Methods
 Ability to apply specific statistical and regression analysis methods applicable to predictive
analytics to identify new trends and patterns, uncover relationships, create forecasts, predict
likelihoods, and test predictive hypotheses.
 Ability to develop and use various quantitative and classification predictive models based on
various regression and decision tree methods.
 Competency-2: Predictive Analytics Tools
 Develop familiarity with popular tools and software used in industry for predictive analytics,
especially R, R Studio and R Markdown.
 Competency-3: The Predictive Analytics Cycle
 Understanding of how to formulate predictive analytics questions.
 Learn how to select the appropriate method for predictive analysis, and how to build effective
predictive models.
 Learn how to search, identify, gather and pre-process data for the analysis.
 Learn how to evaluate the soundness, appropriateness and validity of their models and how to
interpret and report on results for a management audience.
Student Requirements and Responsibilities

Last updated: 5/6/2016

2
 Students need to be familiar with this syllabus and the weekly class schedule below. All assignments
and class events will be posted either in the class schedule or on Blackboard.
 Similarly, students need to check all announcements posted on Blackboard before each class.
 Students are required to check their American University e-mail regularly for class announcements.
Students who do not use their AU e-mail regularly need to either forward their AU e-mail to their
personal e-mail accounts or change their e-mail address in Blackboard.
 Students are required, per University policy, to be familiar with AU's Academic Integrity Policy.
Please read carefully the policies and read the Academic Integrity Policy section below. These
policies will be strictly enforced in this course.
 Students are required to read all assigned material prior to class, prepare for class as instructed,
participate actively in class discussion, and take a proactive role to maximize their learning from this
class and in helping others benefit from the course. Students must read the assigned material
before class and review the R code and related instructions before the corresponding R sessions.
 A good portion of the class lectures will come from sources other than the textbooks. Therefore,
this class requires regular attendance and consistent week-to-week commitment on the part of the
student. The material in this course is sequential in nature, so missing a lecture will not only affect
the student's learning on the missed lecture, but also on subsequent material covered.
Grading Structure

Course Component Weight Composition

4 Homework 20% (4 @ 5% each) Individual
Exam 30% Individual
Term project 30% Individual or Team
Quizzes and class exercises 10% Individual
Attendance and participation 10% Individual
Total 100%

Grading Legend:
A: 93 or above; A-: 90 to less than 93;
B+: 88 to less than 90; B: 83 to less than 88; B-: 80 to less than 83;
C+: 78 to less than 80; C: 73 to less than 78; C-: 70 to less than 73;
D: C-: 60 to less than 70;
F: less than 60.
Course Components (all work in this course is individual)
1. Homework: 4 homework assignments on predictive analytics modeling. The homework will focus
on hands-on use of R software to develop predictive models. The homework will be prepared in R
Markdown and submitted as an HTML file (Produced by knitr HTML from R Markdown).

2. Exam: There one in-class exam towards the end of the semester. The exam will be conceptual. The
exams will aim at testing your ability to process various business scenarios/problems/questions and

Last updated: 5/6/2016

3
 select and justify specific predictive modeling method. The exam covers all lectures and ISLR
textbook readings up to and including the last class. R coding will not be covered in the exam but
you need to be able to interpret plots and other outputs I prepared in R. In each question you will be
presented with an analytics scenario. This scenario will contain one or more of the following:

2.1. A particular problem to resolve or business question to answer with predictive analytics
(important note: your goal will NOT be to solve the problem or answer discussion, but to
discuss your approach to do that);

2.2. An analysis goal (i.e., interpretation, inference or prediction); and

2.3. Relevant exhibits, which may include things like: model summary outputs; plots; distributions;
data descriptions or displays. I will not ask any questions on R coding, but I will include a few R
plots and outputs for interpretation. Each question will require a short, concise and precise
answer demonstrating your knowledge of the material understanding of the specifics of the
scenario, rather than long essays with vague generalities.

3. Term Project: The project will be done in teams of maximum 3 students. Students are also
welcomed to work in pairs or even individually. For students working in teams, it is expected that all
team members will contribute equally and that everyone will take the opportunity to learn from
each other. Students will identify a business problem to address through predictive analytics. The
goal is to select appropriate models and model specifications, and apply the respective methods to
enhance data-driven decision making related to the business problem. Students will identify
potential use of predictive analytics, formulate the problem, identify the right sources of data,
analyze data, and prescribe actions to improve not only the process of decision making but also the
outcome of decisions. See further instructions on Blackboard.

4. Quizzes and Class Exercises: You will complete several quizzes and class exercises during the
semester. The quizzes will be short (10 minutes or so) based either on assigned material or on
material already covered in class. The class exercises will involve short R assignments in class. There
will be 8 to 10 quizzes and/or exercises during the semester. The lowest grade will be dropped from
your average. So try not to miss more than one of these.

5. Class attendance, participation and exercises: Attendance is a straight percentage of the classes
you attended, adjusted for lateness and early departures; In-class participation is measured by the
ability of students to bring quality discussion into the class. This course is based on a model of
active learning, with class discussions and exercises playing a central role. Students are expected to
read the assigned material and to carefully prepare for all cases and exercises before coming to class
and completing the required class exercises, when assigned. Students will be called upon to respond

Last updated: 5/6/2016

4
 to faculty questions. This course is very hands-on and the only way to learn the material well is
through intensive exercises. About 50% of the class will be focused on hands on demonstrations and
graded exercises.

Class Schedule

Note: The textbook authors have a nice series of video lectures where they narrate the book themselves.
While these videos are not required, they really help understand the readings. You can find all the video
lectures associated with each chapter at:
http://auapps.american.edu/~alberto/analytics/ISLRLectures.html
Activity
Student Learning (R) Readings; (HW) Homework;
Week Topic
Objectives (E) Exam; (P) Project; (V) Watch
Video Lecture; (o) Other
1A Develop a deep 1. Introduction All authors’ slides and video
Tu understanding of the  Syllabus Overview lectures available at:
5/16 predictive analytics life  Course Introduction http://auapps.american.e
cycle and several  The Analytics Life Cycle du/alberto/www/analytic
foundational concept that  Introduction to Predictive Analytics s/ISLRLectures.html
will be used throughout  Matrix Notation (R) ISLR Ch.1 Introduction
the course.  Basic Foundations (V) ISLR Ch.1
 (B) Model, Method and Feature (O) Download and install R
Selection and R Studio

1B Overview of R for 2. R Refresher (R) ISLR 2.3 Lab: Introduction

Th Predictive Modeling to R
5/18 3. Regression Refresher (O) Recommended R book: R
Overview of basic  Covariance, Correlation and ANOVA for Everyone
statistics and the review.
Ordinary Least Squares  Simple Linear Regression
(OLS) regression model  OLS Model Diagnostics

2A Further insights into the 3. Regression Refresher (cont’d.) (R) ISLR Ch.3 Linear
Tu OLS regression model and  Dummy Variables Regression
5/23 its assumptions and  Multivariate Regression
limitations. Exploring the  OLS Assumptions HW1 Due
first departure from OLS  Weighted Least Squares (WLS) R Practice
due to heteroscedasticity  Generalized Linear Models (GLM)
WLS. Taking a first look at
GLM

Last updated: 5/6/2016

5
 2B Learning to work with 4. Data Pre-Processing
Th various data types and  Overview
5/25 how to pre-process the  Variable Types
data for analysis,  Introduction to Data Transformations
including popular  Data Transformations:
transformations like Box- 1. Categorical to Dummy Variables
Cox, standardized data, 2. Polynomials
log transformations and 3. Box-Cox Transformation
lagging time series data.

3A 4. Data Pre-Processing (cont.d) (R) ISLR Ch.2 Statistical

Tu 4. A) Log & Elasticity Models Learning
5/30 B) Logit Transformation (R) ISLR 5.1 Cross-Validation
5. Count Data Models
6. Centering
7. Standardization

3B 4. Data Pre-Processing (cont.d)

Th 8. Rank Transformations
6/1 9. Lagging Data (Causal Models) HW2 Due
10. Data Reduction

4A Learning the basic 5. Machine Learning

Tu concepts behind  Machine Learning Overview
6/6 “machine learning” and  Bias vs. Variance Tradeoff
the various ways of  Error Measures
evaluating the predictive  Cross-Validation
accuracy of models.

4B Learning how to select 6. Variable Selection (R) ISLR Ch.6.1 Linear Model
Th the number of predictors  Dimensionality Issues Selection
6/8 in a model and address  Multi-Collinearity
issues of dimensionality,  Variable Selection Methods
like multi-collinearity.  Step Methods HW3 Due

5A Learn how to build 7. Non-Linear Models (R) ISLR Ch.7 Beyond Linearity
Tu predictive models when  Non-Linearity Overview
6/13 the relationship between  Interaction Models
the predictors and the  Polynomial Models (P) Term Project
outcome variable don’t  Step Models Proposal Due
appear to follow a linear  Piecewise Models
trend.  Piecewise Linear Models
 Piecewise Polynomial Models
 Spline (MARS) Models

Last updated: 5/6/2016

6
 5B Learn how to build 8. Classification Models (R) ISLR Ch.4 Classification
Th predictive models when  Introduction
6/15 the outcome is binary  Binomial Logistic Regression
(e.g., yes/no,  Multinomial Logistic Regression HW4 Due
success/failure,  Linear Discriminant Analysis
approve/decline), using  Quadratic Discriminant Analysis
popular methods like
logistic regression and
discriminant analysis.

6A Learn the various 10. Decision Trees (R) ISLR Ch.8 Tree-Based
Tu methods to build  Decision Trees Methods
6/20 predictive classification  Regression Trees
models using decision  Growing Trees
trees, rather than  Regression Tree Issues
regression models.  Classification Trees
 Pruning Trees
 Bootstrap Aggregation (Bagging)
 Random Forest Models

Review for Exam

6B
Th Exam (up to and including 8 above)
6/22

7A Learning how to handle 9. Dimensionality (R) ISLR Ch.6.2 Shrinkage

Tu predictive with large  (D) Regularization (Penalized or Methods
6/27 number of predictors, Shrinkage Models)
and how to reduce the  Ridge Regression
set of predictors using  LASSO
regularization, penalized  (D) Dimension Reduction Models
models and other  Principal Components Regression
dimension reduction (PCR)
methods like principal  Partial Least Squares (PLS)
components and partial
least squares. Model/Method Selection Review
Fly Solo
Reviews for Term Project
Course Wrap-Up

7B
Project Reports in Class (P) Term Project Due
Th
IN CLASS
6/29

Last updated: 5/6/2016

7
Academic Integrity Code
Academic integrity is paramount in higher education and essential to effective teaching and learning.
As a professional school, the Kogod School of Business is committed to preparing our students and
graduates to value the notion of integrity. In fact, no issue at American University is more serious or
addressed with greater severity than a breach of academic integrity.
Standards of academic conduct are governed by the University’s Academic Integrity Code. By enrolling
in the School and registering for this course, you acknowledge your familiarity with the Code and pledge
to abide by it. All suspected violations of the Code will be immediately referred to the Office of the
Dean. Disciplinary action, including failure for the course, suspension, or dismissal, may result.

Additional information about the Code (i.e. acceptable forms of collaboration, definitions of plagiarism,
use of sources including the Internet, and the adjudication process) can be found in a number of places
including the University’s Academic Regulations, Student Handbook, and website at
<http://www.american.edu/academics/integrity>. If you have any questions about academic integrity
issues or about standards of conduct in this course, please discuss them with your instructor.

Academic Support Services

If you experience difficulty in this course for any reason, please don’t hesitate to consult with me. In
addition to the resources of the department, a wide range of services is available to support you in your
efforts to meet the course requirements.

Academic Support Center (x3360, MGC 243) offers study skills workshops, individual instruction, tutor
referrals, and services for students with learning disabilities. Writing support is available in the ASC
Writing Lab or in the Writing Center, Battelle 228.

Counseling Center (x3500, MGC 214) offers counseling and consultations regarding personal concerns,
self-help information, and connections to off-campus mental health resources.

Disability Support Services (x3315, MGC 206) offers technical and practical support and assistance with
accommodations for students with physical, medical, or psychological disabilities. If you qualify for
accommodations because of a disability, please notify me in a timely manner with a letter from the
Academic Support Center or Disability Support Services so that we can make arrangements to address
your needs.

Kogod Center for Business Communications (x1920, KSB 101) To improve your writing, public speaking,
and team assignments for this class, contact the Kogod Center for Business Communications. You can
get advice for any written or oral assignment or for any type of business communication, including
memos, reports, individual and team presentations, and PowerPoint slides. Hours are flexible and
include evenings. Go to http://www.kogod.american.edu/cbc and click on "make an appointment," visit
KSB 101, or email [email protected]. You may also call x1920.

Last updated: 5/6/2016

8
Financial Services and Information Technology Lab (FSIT) (x1904, KSB T51) to excel in your course work
and to maximize your business information literacy in preparation for your chosen career paths, we
strongly recommend to take advantage of all software applications, databases and workshops in the
FSIT Lab. The FSIT Lab promotes action-based learning through the use of real time market data and
analytical tools used by business professionals in the market place. These include Bloomberg, Thomson
Reuters, Argus Commercial Real Estate, Compustat, CRSP, @Risk etc. For more information, please
check out the website at Kogod.american.edu/fsit/ or send us an email to [email protected].

EMERGENCY PREPAREDNESS FOR DISRUPTION OF CLASSES

In the event of an emergency, American University will implement a plan for meeting the needs of all
members of the university community. Should the university be required to close for a period of time,
we are committed to ensuring that all aspects of our educational programs will be delivered to our
students. These may include altering and extending the duration of the traditional term schedule to
complete essential instruction in the traditional format and/or use of distance instructional methods.
Specific strategies will vary from class to class, depending on the format of the course and the timing of
the emergency. Faculty will communicate class-specific information to students via AU e-mail and
Blackboard, while students must inform their faculty immediately of any absence. Students are
responsible for checking their AU e-mail regularly and keeping themselves informed of emergencies. In
the event of an emergency, students should refer to the AU Student Portal, the AU Web site
(http://www.american.edu/emergency/) and the AU information line at (202) 885-1100 for general
university-wide information, as well as contact their faculty and/or respective dean’s office for course
and school/ college-specific information.

Last updated: 5/6/2016

9