Stats 300C: Lectures







  • Lecture 1: Global testing, Bonferroni's global test, Fisher's combination test
  • Lecture 2: Global testing, optimality of Bonferroni's method for single strong effect
  • Lecture 3: Global testing, chi-square test, optimality of chi-square test for distributed mild effects
  • Lecture 4: Global testing, Simes test, Tests based on the empirical cumulative distribution: Kolmogorov-Smirnov, Cramer-von Mises, Anderson-Darling and Tukey's second-level significance testing
  • Lecture 5: Multiple testing/comparison problems; FWER, procedures for controlling FWER, Holm's step-downn procedure
  • Lecture 6: Multiple testing/comparison problems; closure principle, step-up and step-down procedures, Hochberg's procedure.
  • Lecture 7: Multiple testing/comparison problems; false discovery rate (FDR), properties of FDR, procedure for controlling FDR (BH(q))
  • Lecture 8: FDR, empirical process viewpoint on Benjamini-Hochberg procedure, Storey's correction
  • Lecture 9: FDR control under dependence.
  • Lecture 10: FDR control under positive dependence, the PRDS property
  • Lecture 11: Multiple testing/comparison problems; empirical Bayes approach to multiple testing; empirical Bayes interpretation of BH(q)
  • Lecture 12: Knockoff filter, FDR control under correlations
  • Lecture 13: Model-free inference and selection, model-free knockoffs
  • Lecture 14: Model-free inference and knockoffs construction for genome-wide association studies (GWAS)
  • Lecture 15: Selective inference, False coverage rate (FCR), FCR-adjusted confidence intervals.
  • Lecture 16: Selective inference: post-selection inference (POSI), inference after model selection via the lasso
  • Lecture 17: Selective hypothesis tests, verifying the winner
  • Lecture 18: Estimation of a multivariate normal mean; Stein's phenomenon, James-Stein estimate, Stein's unbiased risk estimate
  • Lecture 19: Estimation of a multivariate normal mean; Empirical Bayes Interpretation of James-Stein
  • Lecture 20: Prediction error, Stein's unbiaded risk estimate, Cp statistic
  • Lecture 21: Model selection with Cp
  • Lecture 22: SURE and prediction error for the lasso
  • Lecture 23: Ideal risk, oracles and risk inflation criterion
  • Lecture 24: Risk inflation of thresholding rules, excess risk of the Lasso
  • Lecture 25: Adapting to unkmown sparsity: FDR thresholding, Sorted L1 penalized estimation (SLOPE)

Last year lectures: 2016 Lectures