789 syllabus

George Mason University
School of Information Technology and Engineering
Department of Applied and Engineering Statistics

STAT 789: Advanced Topics in Statistics (Topics in Applied Statistics)

Summer Session, 2002
Tuesdays and Thursdays from 7:20 to 10:00 PM

starting May 28 and ending July 23
with no class meetings on July 4 and July 18

Location: room B120 of Robinson Hall (note: Robinson has an A wing and a B wing)

Instructor: Clifton D. Sutton

Contact Information (phone, fax, e-mail, etc.)
Office Hours: 6:00-7:00 & 10:00-10:30 PM Tuesdays and Thursdays (more information)

Text:

Beyond ANOVA: Basics of Applied Statistics, Reissue edition, by R. G. Miller, Jr. (Chapman & Hall, 1996)

Prerequisite:

Permission of instructor (but if a student has taken STAT 554 at GMU and is currently in an M.S. or Ph.D. program, there is no need to contact me before registering)

Description:

This course will cover an assortment of topics that either "fall between the cracks" of the regular graduate-level statistics classes at GMU, or may be briefly mentioned in some courses, but not covered with sufficient depth. The topics addressed this summer will primarily be selected from the outline below. (I doubt that there will be sufficient time to include all of the topics indicated.)

Items 1 through 6 below may constitute roughly half of the course, with items 7 and 8 receiving the remaining time (although I will alter this plan as appropriate as the details of the course are firmed up throughout the summer). (Here is an updated outline of what the course actually covered.)

Single sample situations
- bootstrap methods (estimates of bias and standard error, confidence intervals)
- jackknifing
- robust estimation (trimmed means, M-estimators)
- dealing with dependent observations (detection and correction)
- density estimation
- permutation test for a matched-pairs experiment
Two sample situations
- permutation test for two independent samples
- robust and bootstrap procedures
K sample situations
- dealing with heteroscedasticity
- comparisons with a control (the many-one problem)
- monotone alternatives (the Abelson-Tukey test, nonparametric alternatives)
Two-factor experimental design situations
- monotone alternatives
- dealing with slightly unbalanced designs (missing values)
Inferences about variances (robust procedures will be emphasized)
- single sample setting
- two sample setting
- K sample setting
Inferences about ratios (the jackknife will be revisited)
Regression
- a brief presentation on OLS regression (including a method for finding good transformations for predictors)
- robust regression (M-estimation (including L_p (including LAD)) and associated tests of parameters, other robust methods)
- the error-in-variables model
- nonlinear regression
Computer-intensive methods for classification and regression
- overview of supervised classification (the Bayes classifier)
- kernel-based and nearest neighbor methods
- the use of test samples and cross-validation for model selection
- classification and regression trees (CART)
- multivariate adaptive regression splines (MARS)

(Here is an updated outline of what the course actually covered.)
For the most part, the plan is to expose students to many of these topics and not get too bogged down in the details. Due to lack of time in the summer session to become familiar with new software, the high cost of some of the pertinent software, and the absence of easy access to some software on campus, you won't be expected to learn to use software to implement all of the methods presented in class --- rather, the chief goal will be to get you to understand the main ideas behind the methods. (If you understand what the methods are about, and when it may be appropriate to use them, then you should be somewhat prepared to make use of them in the future when you have easy access to suitable software and have time to learn to use the software.) I will present results obtained using a variety of software, and will occasionally give demonstrations. For some portions of the course, you will be expected to use appropriate software to analyze data.

Grading:

50% for homework assignments
- (more information can be found here)
25% for quizzes (part of the 1st quiz will be closed books and notes, but all other quizzes will be open books and notes)
- (more information can be found here)
25% for final exam (open books and notes)
- (more information can be found here)

Additional Comments:

I recommend that students bring copies of the STAT 554 lecture notes to class, since I will feel free to refer to them during lectures
be sure to note that there is no class on the 4th of July, nor on July 18
I can possibly make arrangements to meet with you outside of my scheduled hours; but on Tuesdays and Thursdays before class I do not like to be bothered from 7:00 to 7:17
please do not leave long messages on my voice-mail, and since I often do not get around to returning calls until the evening, you should state what time you plan to go to sleep (and always leave both your day and evening phone numbers, speaking slowly, even though you may have given them to me previously); I much prefer to communicate in person or via e-mail than over the phone (phone tag gets frustrating, and I've experienced problems with the reliability of the GMU voice-mail system --- but I will try to return your calls if you're unable to communicate via e-mail when you're off campus)
please abide by the university policy that cell phone ringers be turned off while class is in session
you are expected to familiarize yourself with the George Mason University honor code and abide by it; although it is perfectly okay to seek assistance from others on most of the homework problems, it will be considered to be a violation of the honor code if you give or receive unauthorized aid on certain specified homework problems and on all quizzes and on the final exam
you are expected to take the final exam during the designated time slot; Incompletes will not be granted except under very unusual circumstances
I will drop at least a couple of low homework scores and at least a couple of low quiz scores --- make-up quizzes will not be given (you have to be in class at the time a quiz is given in order to receive any credit for it)
any class meetings canceled by the university due to snow, sleet, power outage, bombing, etc. will be made up if possible; with regard to bad weather, I plan to teach class if the university is open and not teach it if the university is closed, so instead of calling me if it snows, simply check to determine if the university is open or closed
caveat: the schedule and procedures described here for this course are subject to change (it is the responsibility of students to attend all class meetings and keep themselves informed of any changes)