Lecture 2 (Wed.., Jan. 17): Consistency of jackkife variance estimator.
HW Assignment 1: Problems 8.4, 8.5, 8.7, 8.9 from Ch. 8. Due Wed., Jan. 24.
Lecture 3 (Mon., Jan. 22): Computing the jackknife in R and using it with simulation studies. Various R codes.
Lecture 4 (Wed., Jan. 24): The jackknife in non-iid situations.
HW Assignment 2: Use the R program to analyze how well the jackkife variance estimator works for one estimator and one distribution for 3 samples sizes, n=10, n=20, and n=200. For the estimator, choose anything you like: trimmed mean, winsorized mean, a sample moment or ratio of sample moments like skew or kurt, etc. For the distribution, anything available in R is fine: rnorm, rt, rgamma, rexp, etc. Due Wed., Jan. 31.
Lecture 5 (Mon., Jan. 29): Begin bootstrap, Ch. 9.
Lecture 6 (Wed., Jan. 31): Bootstrap standard errors, consistency, choosing B.
HW Assignment 3: 9.3, 9.4. Due Wed., Feb. 7.
Lecture 7 (Mon., Feb. 5): Bootstrap confidence intervals.
Lecture 8 (Wed., Feb. 7): Confidence intervals continued.
HW Assignment 4: Use the same Monte Carlo data and estimator as in HW 2. Repeat HW 2 using the bootstrap in place of the jackknife (boot.var) to get the variance estimates (but don't change the function var.est that analyzes the output). Then put the results from the jackknife and bootstrap into one table with average se's similar to Table 9.1. Due Wed., Feb. 14.
Lecture 9 (Mon., Feb. 12): Bootstrap testing. R code for power estimates based on B=99 bootstrap replications.
Lecture 10 (Wed., Feb. 14): Permutation procedures introduction.
Lecture 11 (Mon., Feb. 19): Theory of permutation tests.
Lecture 12 (Wed., Feb. 21): Various forms of the Wilcoxon Rank Sum statistic and approximations. Sample code for Wilcoxon Rank statistics
HW Assignment 5: 10.1 and calculate E(W) from the distribution and compare to n(n+m+1)/2, 10.2, 10.6, 10.12. Due Wed., Feb. 28.
Lecture 13 (Mon., Feb. 26): Box-Andersen approximation, Locally Most Powerful rank Tests.
Lecture 14 (Wed., Feb. 28):Pitman ARE, k-sample problem, two-way setup.
Spring Break
Lecture 15 (Mon., Mar. 12): Model selection overview, variable selection for linear regression, MSE of prediction calculations.
Lecture 16 (Wed., Mar. 14): Mallows Cp and other linear model criteria. R program and example output for forward addition sequence. R program and example output for backward elimination sequence.
Class Project:Project Instructions
Lecture 17 (Mon., Mar. 19): Information criteria: AIC, BIC, GIC, etc. Asymptotic consistency in variable selection.
Lecture 18 (Wed., Mar. 21): Cross-validation, bootstrap, etc., in variable selection.
Lectures 19-22 (Mar. 26, 28, Apr. 2, 4 ): Information-Reduction/Noise-Addition methods in variable selection. Lecturer: Dr. Stefanski.
Boos-Stefanski Variable Selection Website
Lectures 23-24 (Apr. 9, 11): Shrinkage/Penalty methods (LASSO, SCAD, OSCAR). Lecturer: Howard Bondell.
Lecture 25 (Mon., Apr. 16): Large numbers of independent variables?
Lecture 26 (Wed., Apr. 18): Bayesian methods in variable selection. Lecturer: Sujit Ghosh. Background paper by Berger and Perrichi. Background paper by Chipman, George, and McCulloch.
Lecture 27 (Mon., Apr. 23): Not sure yet.
Lecture 28 (Wed., Apr. 25): Project presentations.