Welcome to the course home page for BST 764: Applied Statistical Modeling for Medicine and Public Health. Announcements and corrections to the notes appear below; links to notes, assignments, and data sets are on the left.
Announcements:
11-10: Assignment 4 is now due on November 22, not November 17.
10-27: Please use the data set prostate2 instead of prostate for problem 6 on assignment 3.
9-8: Assignment 1 is now due on September 20, not September 15.
8-7: Course webpage created.
Changes/corrections to notes/assignments:
11-16: Two corrections to the 11-15 notes. Slide 10, first bullet: "eigenvalues" should be "eigenvectors"; slide 16, third bullet: V should be VT.
11-10: A few corrections to Assignment 4: xj should be xi in the definition of RSSj; all less than signs should be less than or equal to signs; for clarification, by "prediction error", I mean "mean squared prediction error". All changes are incorporated in the current online copy of Assignment 4.
10-27: Data set prostate2.txt added, which modifies the original (removing some transformations) and fixes a data entry mistake
10-27: 10-27 notes, slide 8: π(-1) should be π(-i)
10-27: 10-27 notes, slide 2: f(x0) should be f(xi)
10-27: 10-20 notes, slide 9: 1-t2 should be 1-u2
9-30: 9-27 notes, slide 28: R code describing how predict.bs works edited for clarity and correctness (newdata should be newx)
9-28: Assignment 2, problem 4: Xjk should be xij|G=k
9-28: 9-15 notes, slide 6, bottom equation; k should be K
9-15: 9-15 notes, slide 3; Sigma was missing a subscript (k)
9-15: 9-15 notes, slide 4; lemma neglected to state that A must be a symmetric matrix
9-15: 9-15 notes, slide 6; bottom equation requires a double summation; only one summation sign was present
9-8: 9-1 notes, slide 7, second equation: Beta changed to bold, as it is a vector.
9-6: Modified lasso definition (slide 3, 9-6 notes), adding a factor of 1/2 so that the definition matches up with the solution derived on slide 7.
9-1: Added a slide (page 12) on AIC and BIC to 9-1 notes.