STAT 100C: Linear Models

Spring 2023

Theory of linear models, with emphasis on matrix approach to linear regression and connections to multivariate normal distribution. Topics include simple and multiple linear regression, model fitting, inference about parameters, testing general linear hypotheses, specification issues, model checking and model selection.

General info

Lectures: TR 9:30am-10:45am @ WGYOUNG 2200
Instructor: Arash A. Amini
Office Hours: No office hours.
TA: Ian McGovern
- TA Office hours: TR 2PM @ Boelter 9406
- TA Discussion Section: M 4pm-4:50pm (DIS 2A) & M 5pm-5:50pm (DIS 2B) @ MS 5128
Grader: Yasi Zhang
Midterm: The midterm will be available on May 12 at 12:00 PM and should be completed no later than May 13 at 6:00 PM. Once you start the exam, you will have 180 minutes to finish it. The exam will be posted on Gradescope, and you will need to submit your completed exam on the platform.
Final Exam: Monday, June 12, 2023, 11:30 AM - 2:30 PM.
Announcements: Will be posted on Campuswire (Sign-up code: 9007)
Homework Submission: Gradescope. (Sign-up code: BBWB8V)
Attendance: iClicker. Sign-up by following this link.
Grading: Attendance 5%, Homework 20%, Midterm 30%, Final 45%.
Prerequisites: STATS 100B and linear algebra such as Math 33A.

Please read!

Please do not email me your late homework. Instead, post a note on Campuswire (it can have attachments) and set the visibility to TAs and Instructors only. We will address the issue there. Only requests through Campuswire are considered.

Resources

Homework, slides and notes are in the Box folder.
Lecture videos are posted in this Youtube playlist.

Past lectures are available here:

Previously recorded lectures -- Winter 2022
Previously recorded lectures -- Spring 2021. (Youtube playlist)
Previously recorded lectures -- Winter 2021 (Youtube playlist)
Previously recorded lectures -- Spring 2020 (Youtube playlist)

Textbook

B. Abraham and J. Ledolter, Introduction to Regression Modeling, 2006. ISBN: 978-0534420758 Corrections, courtesy of Prof. Ledolter.

Data

Datasets for the book.

Supplementary texts

J. J. Faraway, Practical Regression and Anova using R: Introduction to doing regression in R.
G. Strang, The Four Fundamental Subspaces: 4 Lines: Short overview of linear algebra.

Syllabus

Lecture	Date	Topic
1	04/04	Review of linear algebra: linear independence; span; basis;
2	04/06	Review of linear algebra: image and column space; rank; inner product; orthogonal complement; projection
3	04/11	Review of linear algebra: spectral decomposition; PSD matrices; Expectation of random vectors
4	04/13	Covariance matrix
5	04/18	Multivariate normal distribution
6	04/20	Linear model; matrix-vector repr.; assumptions of the model; MLE of params.
7	04/25	MLE of params; geometric intepretation; variance estimation
8	04/27	Statistical properties of estimators; hat matrix; projection matrices
9	05/02	Inference I: Review of hypothesis testing (HT) and confidence intervals (CI)
10	05/04	Inference II: HT and CI for scalar parameters in linear models
11	05/09	Inference III: CI for regression function; prediction intervals
12	05/11	Inference IV: General linear hypothesis; geometrice intepretation
13	05/16	Inference V: Additional sum of squares and the F-test
14	05/18	Inference VI: Comparing models; ANOVA; R^2 and overfitting
15	05/23	Comparison of treatments (breif); qudratic forms and their distributions
16	05/25	Cochran's theorem; proof of F-test theorem; Gauss-Markov theorem
17	05/30	Generalized least-squares (GLS); Multicoliearity; variance inflation factor (VIF)
18	06/01	Diagnostics; Oultier detection: leverage and influence
19	06/06	Model selection
20	06/08	Modern approaches; advanced topics

Miscellaneous

For statistical computation, R is recommended.

p-value controversies: