INTEGRATED PROGRAMME IN MANAGEMENT (IPM)

AY2023-24 BATCH: 2022-27 TERM: IV

TITLE OF THE COURSE: STATISTICAL METHODS - II

CREDITS: 6

Name of the Faculty Faculty Email Telephone

Block/ Number
Room No.
Sudeep R. Bapat B-212 [email protected] +91-8377881090

CONSULTATION TIME FOR STUDENTS

Name of the Faculty Timing
Sudeep R. Bapat By prior appointment

COURSE DESCRIPTION

This course presents advanced methods in Statistics, which follows topics covered during
the first year. The topics broadly include simple and multiple linear regression. We will also
focus briefly on categorical data, and binary responses, which use a logistic regression.

COURSE OBJECTIVES

1) Statistical data analysis for students in management sciences

2) Analyzing statistical data using R

3) Simple and multiple linear regression with applications

4) Introduction to logistic regression with examples

PEDAGOGY/TEACHING METHOD:

Mix of lectures, R sessions, classroom-problem solving sessions, homeworks and exams.

Expected Learning Outcomes and Associated Measures

At the end of the course, the student is expected to accomplish the following course learning
outcomes (CLO). Alignment of CLO with the Program Level Goals & Objectives and
Assessment of the learning outcomes of the course is presented below.

Course Learning Outcome Program Level Goals/ Assessment Tool(s)

Outcome

To enable students to PLG 3. Demonstrates Embedded question(s) in

understand the concepts of knowledge of basic Quiz(s)/Mid-Term/End-
Mathematical Statistics, and foundations of Management Term Examinations.
to apply these techniques to
Studies
real world applications and
scenarios. 2.1. Acquires analytical
rigour through Mathematics
and Statistics

TEXTBOOK FOR THE COURSE

Kutner, M., Nachtsheim, c., Neter, J. and Li, W. (2004), Applied Linear Statistical Models,
5th edition, McGraw-Hill/Irwin, New York, USA. [ALSM]

Note: The instructor will provide additional course notes.

EVALUATION

Individual Component Group Component Weightage

Homeworks 20
Quizzes 15
Mid-term examination 25
End-term examination 40
Total 100%

ACADEMIC DISHONESTY

IIM Indore believes in Academic honesty.

Academic dishonesty or misconduct is cheating that relates to an academic activity. It is
a violation of trust between the Institute and its stakeholders. Plagiarism, fabrication,
deception, cheating, and sabotage are examples of unacceptable academic conduct. Please
consult the Program manual for the section on academic dishonesty.

SCHEDULE OF SESSIONS

Module I: Simple Linear Regression (SLR)

Module Objective: Introduction to simple linear regression and associated inference

Session 1 - 2 SLR model, assumptions, model fitting in R, inference related to model

parameters, ANOVA table, data visualization.

Objective: Introduce the concept of simple linear regression.

Reading: Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 1: Pages 2-13

Session 3-4 Parameter estimation via OLS and MLE. Usage of lm() and optim()

Objective: Discuss the theory and implementation (via R) of parameter estimation.

Finding parameter estimates using other optimization tools like BFGS,
brute-force grid search or genetic algorithm.

Reading: Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 1: Pages 13-25

Session 5 Practice Problems

Objective: Hands on exercises for both theoretical and R-based questions.

Reading: As given during that week

Session 6 Predictive analysis, confidence vs prediction intervals

Objective: Discuss both point and interval estimates of the mean response and
prediction of new observation at a specific input. Derive the closed form
expressions, and show the usage of predict.lm().

Reading: Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 2: Pages 40-60
 Session 7 - 8 Diagnostics and Residual analysis

Objective: Understanding the significance of different model assumptions

(including outliers and homoscedasticity). Introduction via theory and
plots.

Reading: Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 3: Pages 100-113
Session 9-10 Practice problems

Objective: Hands on exercises for both theoretical and R-based problems

Reading: As given during that week

Module II: Matrix representation of SLR and multivariate normal distribution

Module Objective: Develop concepts such as random vectors and multivariate data
analysis with focus on MVN

Session 11 Matrix representation of SLR model, bivariate normal, multivariate

normal, conditional normal

Objective: Introduce concepts of random vector and multivariate data, mean vector
and covariance matrix, multivariate probability densities, marginal and
conditional densities

Reading: Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 5: Pages 176-196
Session 12-13 SLR model, estimation, inference, and diagnostics

Objective: Introduce the general form of SLR model in a matrix form to provide a
foundation for MLR model

Reading: Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 5: Pages 197-203
Session 14 SLR inference, quadratic forms, and Chi-square distribution

Objective: Discuss the analysis of regression models, hat matrix, degrees of

freedom etc. from a matrix notation point of views

Reading: Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 5: Pages 204-212
 Session 15-16 Practice problems

Objective: Hands on exercises for both theoretical and R-based problems

Reading: As given during that week

Module III: Multiple Linear Regression Model (MLR)

Module Objective: Discuss MLR in detail with a thorough understanding of the model
statement, additional assumptions, fitting procedures etc.

Session 17 MLR model statement, general linear model, assumptions, fitting,

results, adjusted R^2

Objective: Introduce model statement, fitting procedure using lm(), parameter

estimates, advantage of matrix notations

Reading: Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 6: Pages 213-225
Session 18-19 Diagnostic checks (linearity and multicollinearity), influential
observations (outliers and high leverage points), partial regression,
sequential ANOVA

Objective: To discuss the validation of additional model assumptions, VIF,

DFbetas, partial correlation coefficients, stepwise regression

Reading: Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 6: Pages 223-230,
Chapter 7: 266-268, Chapter 10: 390-400
Session 20 Polynomial regression and interaction effects

Objective: Introduce polynomial regression via MLR, improvement of model R^2

with interaction effects, geometric interpretation, and visualization of
interaction effects

Reading: Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 8: Pages 294-313
Session 21 Qualitative predictors, piecewise linear regression

Objective: Discuss piecewise models as a special case of MLR with qualitative

predictors

Reading: Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 8: Pages 313-327
 Session 22-23 Model/Variable selection (AIC, Mallow’s Cp, MSPE, PRESS, cross
validation via k-fold and Monte Carlo

Objective: To introduce ranking criteria for model selection and variable selection.
Discuss different goodness of fit measures

Reading: Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 9: Pages 343-368
Session 24-25 Practice problems

Objective: Hands on exercises for both theoretical and R-based problems

Reading: As given during that week

Module IV: Violations in MLR Model Assumptions

Module Objective: To give a brief overview of regression models beyond the standard
MLR models

Session 26 Review of assumptions, Multi-collinearity, Ridge regression, Non-

constant variance, VST, weighted least squares

Objective: To discuss the validation of additional model assumptions and possible

approaches of handling violations. Focus on independence of predictors
and homoscedasticity
Reading:
Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 11: Pages 421-449
Session 27 Non-parametric regression, Lowess, Trees, Non-normality, Box-Cox
transformation, Bootstrap, Autocorrelated errors, Non-linear regression
models (eg. Neural networks)

Objective: Discussing the assumption of parametric relationship between response

and predictors and normality of data. Overview of non-linear regression
models as an alternative to MLR when the linearity assumption fails

Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
Reading: models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 11: 449-460
 Session 28 Generalized linear models

Objective: Discuss logistic regression models (statement, fitting procedures and

inference) under binary response. Briefly discuss the class of GLMs for
categorical response

Reading: Kutner, M., Nachtsheim, C., Neter, J. and Li, W. Applied linear statistical
models, Irwin, USA, McGraw Hill (5th Ed.): Chapter 14: Pages 555-570
Session 29-30 Practice problems

Objective: Hands on exercises for both theoretical and R-based problems

Reading: As given during that week

Please indicate the changes made in the course outline based on the measurement of
assurance of learning (closing the loop)/student feedback:

1. The edition of the book used for lab examples and exercises is upgraded and
mentioned under “Additional Readings”

Please give the details of the book if students need to buy the book
Author Title Publisher Edition Remarks, if
any
Kutner, M., Applied Linear McGrawHill/ 5th
Nachtsheim, C, Statistical Irwin
Neter, J. and Models
Li, W.

Additional Readings (for practical examples using R)

1. Weisberg, S. (2014): Applied Linear Regression, 4th Edition, Wiley.

******