Indian Institute of Management Kashipur

Postgraduate Program in Management

Advanced Machine Learning

Term V, 2019-20
Credit: 1 (Full credit)

Instructors
Name: Prof. Venkataraghavan K
Office: D1-6, Ground Floor, Academic Building, IIM Kashipur
Email: venkat.krishnaswamy@iimkashipur.ac.in
Office Hours: During 1 hour period immediately after the class. No appointments required.

Name: Prof. Mayank Sharma

Office: D1-7, Ground Floor, Academic Building, IIM Kashipur
Email: mayank.sharma@iimkashipur.ac.in
Office Hours: During 1 hour period immediately after the class. No appointments required.

Prerequisites: None
Cross-listed with: IT & Analytics Track

Course Description
Companies have realized the importance of data analytics in decision making. Companies can take path of
data driven decision making based on insights from analytical models.

Important aspect of this course is that it provides an advance technology foundations continuing with
previous courses of analytics that is required for Business Analytics as well as provides cross-functional
application of Business Analytics to marketing, finance, operations, strategy and HR.

This course will provide you with an analytical and technical framework to understand the emerging
world of business analytics from technology and business perspectives.

Course Objectives
At the end of this course, you will:
  PLO1a, PLO1b - Understand recommender systems
 PLO1a, PLO1b - Learn basics of advances topics such as deep learning, boosting, gams
 PLO1a, PLO1b - Learn to develop machine learning models using R and Python
 PLO1f - Learn through hands-on exercises and examples across industries
 PLO1f - Apply and develop analytical models based on business problem(s)
 Learn to communicate the insights derived to stakeholders
 Learn how to leverage business analytics.

Required Textbook
None. Instructor will provide notes as required.

Pedagogy and Course Requirements

The course will be primarily taught though a combination of class discussions, quizzes, term-
long project, and case analysis, and presentations.

Class Discussions

The class discussion will involve the readings assigned for the class, which may include book
chapters, articles, and cases. Students are expected to have done all the assigned reading and to
actively participate in these discussions.

Quizzes/Assignment

N quizzes/assignments will be given at random and without announcement. Each

quiz/assignment will cover the material assigned for and/or discussed in that class. For quizzes if
you are late for class, you will not receive extra time to complete the quiz. The missed
quizzes/assignment shall not be made up under any circumstances and you will receive zero
points.

Group Project

Each team will provide a written report of their project work at the end of the term, but before
the presentation is scheduled.
Each project team will make a presentation of their work in the class. Each member must present
an equitable share of the overall team presentation. Instructor will let you know beforehand the
time allowed for your presentation.

Class participation

Each student will analyze the case problems individually assigned for that class. The analysis
will involve answering questions assigned for the case(s).
 During the class discussion, the student must be prepared to share his/her analysis when called
upon and only when called upon. S/he must move the discussion forward; simply paraphrasing
others’ analysis would earn negative points.

Your class participation grade will reflect evaluation of your participation during class
discussions.

Grading
Component Weightage PLOs
Mid-Term Exam 30% PLO1a/b
End-Term Exam 30% PLO1a/b
Quiz/Assignment 20%
End Term Project 20% PLO1f

Tentative Class Schedule

(Additional readings, cases, and articles may be assigned as they become available)
Session Topics Readings
1 Introduction Case: Netflix Leading With Data: The Emergence
Of Data-Driven Video
https://cb.hbsp.harvard.edu/cbmp/product/KEL473-
PDF-ENG

2 Feature Selection – Part I Conceptual Discussion and Business Applications

 Genetic Algorithms Feature Reduction
 Simulated Annealing  Feature Selection
 Random Forest  Feature Reduction
 Step-wise Regression  Feature Importance
 L1, L2 Regularization

3 Feature Selection – Part II Hands-on Class using (R/Python)

 Genetic Algorithms  Effect of tuning parameters
 Simulated Annealing
 Random Forest
 L1, L2 Regularization
 Random Forest

4 Soft computing methods: Rough Sets  Introduction to Rough Set theory

 Concepts of Set Approximation &
Indiscernibility
 Reducts
 Hands-on class (R/Python)
5 Soft computing methods: Fuzzy Sets  Introduction to Fuzzy set theory
 Concepts of crisp and fuzzy set
 Fuzzy set operations
 Defuzzification
Hands-on class (R/Python)
6 Association Rules – I  Introduction to Association Rules &
Pattern detection
 Key concepts of Support & Confidence
 Other Measures of Interestingness
 Apriori Algorithm
 Application of Association Rules (e.g.
Market Basket Analysis)

7 Association Rules - II Effect of Support & Confidence Values on Rule

Generation. Hands-on class (R/Python)

8 Sequential Data Analytics  Introduction to Sequential Data

 Applications of Sequential Data Analytics
Hands-on class (R/Python)
9 Recommender Systems - I  Introduction to Recommender Systems
 Applications of Recommender Systems
 Item-based collaborative filtering
 User-based collaborative filtering

10 Recommender Systems - II Building & Evaluating Recommender System

Hands-on class (R/Python)
MID TERM
11 Advanced Ensemble Models – Part I Conceptual Discussion and Business Applications
 Bagging Hands-on Class using (R/Python)
 Boosting  Effects of shrinkage
o Adaptive Boosting  Effects of learners
o Gradient Boosting  Loss functions
o Stochastic Gradient Boosting
o XGBoost

12 Advanced Ensemble Models – Part II Hands-on Class using (R/Python)

 Bagging  Effects of shrinkage
 Boosting  Effects of learners
o Adaptive Boosting  Loss functions
o Gradient Boosting
o Stochastic Gradient Boosting
o XGBoost

13 Machine Learning Advanced Linear and Non- Conceptual Discussion and Business Applications
linear Models – Part I  Smoothing Effects
 Generalized Additive Models
o Loess
o Kernel Methods
 o Splines
14 Machine Learning Advanced Linear and Non- Hands-on Class using (R/Python)
linear Models – Part II  Smoothing Effects
 Generalized Additive Models  Parameter tuning/ Grid Search
o Loess  Bsplines, PSplines, REML, REML with
o Kernel Methods Shrinkage
o Splines

15 Introduction to SVM Conceptual Discussion and Hand-on Class using

R/Python
 SVM Kernels
 Grid Search
16 Introduction to Neural Networks Hands-on Class using (R/Python)

17 Deep Learning and Tensor Flow Python/ Keras/ Tensorflow

Guest Lecture I
18 Deep Learning and Tensor Flow Python/ Keras/ Tensorflow
Guest Lecture II
19 Managing Data Science and Machine Learning Industry Best Practices
Projects – Guest Lecture III
20 Student Project Presentation In-class presentations
END TERM EXAM

Course Policies

1. Responsibility for Course Materials:You are responsible for all material covered in
class. If you are absent, you are responsible for obtaining the information you missed.
2. Classroom Behavior: We expect you to participate in class activities in a mature and
appropriate manner. Disruptive or otherwise unacceptable behavior will not be tolerated.
3. Mobile: Mobiles are not permitted in the classroom. I will let you know beforehand if
laptop is required for a class.
4. Academic Conduct:All members of the academic community at IIM Kashipur are
expected to practice and uphold standards of academic integrity and honesty. Academic
integrity means representing oneself and one’s work honestly. Misrepresentation is
cheating since it means students are claiming credit for ideas or work not actually theirs
and are thereby seeking a grade that is not actually earned. Following are some examples
of academic dishonesty:
i. Cheating on quizzes and examinations. This includes using materials such as
books and/or notes when not authorized by the instructor, copying from someone
else’s paper, helping someone else copy work, substituting another’s work as
one’s own, theft of exam copies, or other forms of misconduct on exams.
ii. Plagiarizing the work of others. Plagiarism is using someone else’s work or
ideas without giving that person credit; by doing this students are, in effect,
claiming credit for someone else’s thinking. Whether students have read or heard
the information used, they must document the source of information. When
dealing with written sources, a clear distinction should be made between
quotations (which reproduce information from the source word-for-word within
 quotation marks) and paraphrases (which digest the source of information and
produce it in the student’s own words). Both direct quotations and paraphrases
must be documented. Even if students rephrase, condense or select from another
person’s work, the ideas are still the other person’s, and failure to give credit
constitutes misrepresentation of the student’s actual work and plagiarism of
another’s ideas. Buying a paper or using information from the World Wide Web
or Internet without attribution and handing it in as one’s own work is plagiarism.
iii. Falsifying records or providing misinformation regarding one’s credentials.
iv. Unauthorized collaboration on computer assignments and unauthorized
access to and use of computer programs, including modifying computer files
created by others and representing that work as one’s own.
v. Unless they specifically indicate otherwise, instructors expect individual, unaided
work on homework assignments, exams, lab reports and computer exercises, and
documentation of sources when used. If instructors assign a special project other
than or in addition to exams, such as a research paper, or original essay or a book
review, they intend that work to be completed for that course only. Students must
not submit work completed for a course taken in the past or for a concurrent course
unless they have explicit permission to do so from both faculty members.

Any academic misconduct will automatically result in a failing grade for the class and the
student will be reported to the committee on academic misconduct for further disciplinary
action.

4. Attendance:As far as I am concerned, you are an adult and it is your decision whether or
not you attend class. However, your decision not to attend a class may have negative
consequences for your class grade. (Please consult PGP Participants’ Handbook for this
purpose).
If you decide to attend a class, you must come to the class and take your seat sufficiently
before the beginning of the class time. Under no circumstances you would be allowed in
once the class has started. You are expected to sit through the class unless you have a prior
permission from the instructor to leave the classroom before the end of the class.
5. Late submission: Any late submission beyond the deadline (even by few seconds) will
result in 0 point. Except in case of emergencies, with a doctor's note, any questions
about late submission will not be entertained.
6. Missed exam: There is no make-up for the missed exams unless the student has
discussed and made an arrangement with the instructor for a valid reason beforehand. In
all other instances, the student must produce a valid doctor's note for the day the student
missed the exam. Such doctor's note must be produced in the same week the student missed
the exam.
7. Grade Discussion:It is the student’s responsibility to monitor his or her own grades and
raise any questions s/he may have within one week of the grades assigned.
8. Extra Credit: No Extra credit shall be given to make-up for missed quizzes, assignments,
exams, project, or poor performance in the course.

Learning Accommodations
To provide equal access to the educational programs and opportunities, IIM Kashipur is dedicated to
providing appropriate accommodations to students with documented disabilities such as attention deficit-
hyperactivity disorders, physical disabilities, sensory impairments, and psychiatric disorders in order to
help them achieve their academic and personal potential. These academic accommodations are provided
to students at no cost.

Inclusivity Statement
IIM Kashipur believes that diversity and inclusiveness are essential to excellence in education and
innovation. Our community represents a rich variety of backgrounds, experiences, demographics, and
perspectives. IIM Kashipur is committed to fostering a learning environment where every individual is
respected and engaged. To facilitate a dynamic and inclusive educational experience, we ask all members
of our community to:
• be open to the perspectives of others
• appreciate the uniqueness of their colleagues
• take advantage of the opportunities to learn from each other
• exchange experiences, values, and beliefs
• communicate in a respectful manner
• be aware of the individuals who are marginalized and involve them
• keep confidential discussions private