Controlling FDR While Highlighting Selected Discoveries
Talkat Robust and High-Dimensional Statistics (Simons Institute workshop, 2018) · 2018-10-29..11-02·
summary
Introduces focus BH, a small modification of Benjamini Hochberg that replaces the rejection count with the sum of data-driven weights Uj so FDR is controlled after post hoc filtering. Proves control under PRDS for monotonic filters and gives the first provable guarantee for outer nodes on trees. Shows simulations on fMRI, GWAS clumping and PheWAS ICD trees where focus BH and a permutation variant maintain post-filter FDR while approaching oracle power, whereas naïve BH plus filtering loses control.
AI-written summary · about
chapters (15)
- 0:00Introduction
- 0:25Foundations of Data Science
- 1:02Inductive Data Analysis
- 2:15Global Error
- 4:44Association Studies
- 8:40Cartoon Examples
- 9:15Cartoon Example
- 14:40Prioritization Vector
- 17:52Filter
- 28:18Simulations
- 32:18Power
- 33:03Plot
- 38:22Multiple Filters
- 38:34Conclusions
- 41:20Discussion
related talks
24mHighlights from Current Programs part 2:
29mApplying Randomized Methods for Zonotope Enumeration and...
30mMatrix-free Construction of HSS Representations Using Adaptive Randomized Sampling
08mAn Aligned Rank Transform Procedure for Multifactor Contrast Tests
01h 17mLightning Talks
32mAdventures with Randomized Algebra for Extreme-Scale Signal Processing
56mIntroducing Decision-Making Under Uncertainty to Medical Research:...
31mBayesian Data Analysis of Nonparametric Models in Clojure