Syllabus

Schedule & Location

MW 3:15-4:30, 200-002

Instructor & TAs

Instructor

Jonathan Taylor

• Office: Sequoia Hall #137
• Phone: 723-9230,
• Email
• Office hours: F 12:00-2:00

Email list

The course has an email list that reaches all TAs as well as the professor: stats306b-spr1213-staff@lists.stanford.edu.

All questions should be directed to this email list, rather than TA or the instructor.

Teaching assistants

• Yunjin Choi
  • Office: Sequoia Hall, #208
  • Email
  • Office hours: M 9:00-11:00
• Alexandra Chouldechova
  • Office: Sequoia Hall, #242
  • Email
  • Office hours: W 11:00-1:00

Textbook

• Generalized Linear Models, McCullagh & Nelder. Though this is marked as required, I will not follow it too closely. It is a great reference.

Notes

I will be writing notes as we go, following in part, some of Brad Efron’s notes. Notes will include computer examples, and be written in ipython notebooks. The examples will be both in R and python.

• One parameter exponential families, part I notebook, pdf
• One parameter exponential families, part II notebook, pdf
• Multiparameter exponential families, part I notebook, pdf
• Multiparameter exponential families, part II notebook, pdf
• Multiparameter exponential families, part III notebook, pdf
• Multiparameter exponential families, part IV notebook, pdf
• Processes notebook, pdf
• Quasilikelihood notebook, pdf
• EM algorithm notebook, pdf

Assignments

• Assignment 1, due Wednesday April 17, 2013. From partI, do exercises 1.2, 1.4, 1.8, 1.12, 1.13, 1.19, 1.21. From partII, do exercises 1.2, 1.4, 1.7, 1.9, 1.11.
• Assignment 2, due Monday April 29, 2013. From partI, do exercises 1.5, 1.6, 1.9, 1.10. From partII, do exercises 1.2, 1.3, 1.5, 1.6.
• Assignment 3, due Wednesday May 15, 2013. From partIII, do exercises 1.2, 1.4, 1.7. From partIV, do exercises 1.6, 1.8, 1.10, 1.14, 1.19.
• Assignment 4, due Wednesday June 12, 2013. From quasilikelihood do exercises 1.1, 1.3. From processes do exercises 1.6, 1.12, 1.17. From EM algorithm do exercises 1.1, 1.3.

Ipython profile

I’ve created an ipython profile for the course, that will load some libraries automatically, which I will use in my examples. To use it, clone the git repo with

cd $HOME/.ipython
git clone https://github.com/jonathan-taylor/profile_stats306b.git profile_stats306b

Then, starting the notebook server with

ipython notebook --profile=stats306b

will give you access to the same profile used in executing the code.

Prerequisites

Some familiarity with linear algebra and statistical methods, preferably having taken some of STATS300 sequence.

Topics covered

This is a course on exponential families and generalized linear models. We will cover the following topics (with some subject to change as we go)

• One parameter exponential families
• Multiparameter exponential families
• Generalized linear models
• Curved exponential families
• EM algorithm
• Survival analysis (?)
• Additional topics (?)

Evaluation

• homework (about 5 total); 50%
• final exam (according to Stanford calendar: M 6/10 @ 8:30AM); 50%

Final exam

• Following the Stanford calendar: Monday, June 10 @ 12:15PM.
• If you cannot take the exam at that time and day, then you will have to take this class in a different quarter. Exceptions will only be made due to official university affairs, such as athletic commitments.

R resources

• An Introduction to R
• R for Beginners
• Using R for Introductory Statistics
• Modern Applied Statistics with S
• Practical ANOVA and Regression in R
• simpleR
• Introduction to R
• R Reference Card
• R Manuals
• R Wiki

python resources

• IPython
• Numpy and scipy
• Numpy tutorial
• Python tutorial