St708 prerequisite St 512 Fall 2003 2:35-3:50 T Th room 210 Harrelson *-------------------------------------------------------------* | Instructor: D. A. Dickey 17 Patterson Hall 5-1925 | | e-mail dickey@stat.ncsu.edu | | Office hours: 10:20-11:30 Mon., Wed. | | or by appointment | | | | Grader: Xiao Ni | | Hours: 9:35am-10:35am,Tuesday and Thursday in 116 Withers | *-------------------------------------------------------------* Text: Appled Regression Analysis 2 ed. (Rawlings, Pantula, Dickey) (Springer 1998) Alternate References: Draper and Smith: Applied Regression Analysis Myers, R. H.: CLassical and Modern Regression Analysis Tentative Outline: Week of Topic Pages in text: Aug. 21 Matrices, Projections 37- 56 (review Ch. 1) Aug. 26 Eigenvalues, SVD 57- 68 Sep. 2 Regression as a projection 76- 92 Sep. 9 Quadratic forms 101-135 Sep. 16 Confidence regions 135-148 Sep. 24 Least squares geometry 184-200 (review Ch. 5) ***** Test 1 (Thursday Sept. 25) Sep. 30 Variable Selection 206-231 Oct. 7 Polynomials 235-262 break Oct. 9 Oct. 14 Polynomials Oct. 21 Class variables 269-316 Oct. 28 Diagnostics 341-391 (review Ch. 10) ****** Test 2 (Thursday Oct. 30) Nov. 4 Transformations 397-426 Nov. 11 Collinearity 433-458 Nov. 18 Nonlinear Regression 485-510 (review Ch. 14) Nov. 25 Nonlinear Regression 485-510 ****** Test 3 (Tuesday Nov. 25) Dec. 2 Unbalanced data 546-592 Final Exam 1-4 Thurs. Dec. 11 in our classroom. (exam schedule http://www.ncsu.edu/reg_records/fall_exa.html ) Homework: Homework will be done using SAS. Problems can be accessed from our SICL home page. http://www.stat.ncsu.edu/~st708_info/ I encourage you to work in groups and discuss the homework problems but, of course, to run and write up your own solutions. Homeworks are due on the dates announced. There will be no penalty for being a bit late unless this policy is abused. The due date of the last homework, however, is firm and is the last date that ANY homework will be accepted. Tests: For each test, you can bring one 8 1/2 by 11 sheet of paper with anything you like written on it (both sides). Other than that, tests are closed book. By the last day of class, I will impute an estimated final exam score for each person, based on the test and homework performance to that point. Points so far = .2*(T1+T2+T3)+.15*HW (75 max) Imputed final = (points so far)/3 (25 max) Estimated total points = sum of these (100 points max) Letter grade cutoff lines will be listed. You can accept your grade based on the imputed final exam grade, or you can take the final. The final exam will be comprehensive. You can bring 3 sheets to the final with you. Tests will count 20 percentage points each, the final exam 25% and homeworks 15% for a total of 100 points. If you must miss a test for some valid reason, let me know and I will impute a grade for that test based on your final exam. I do not give makeup exams. Integrity: University regulations require that every course syllabus remind students that the Code of Student Conduct defines a university policy on academic integrity already pledged by each student. Instructors assume that the students' names on their submitted work imply compliance with this policy. See http://www.fis.ncsu.edu/ncsulegal/41.03-codeof.htm Examples: A nice web page with SAS examples from the notes has been set up by Kevin Scott. To view this go to (note caps) http://www.stat.ncsu.edu/~st708_info/dickey/Examples/ST708index.htm