The Knockoff Filter for Controlled Variable Selection
The knockoff filter is a general procedure for controlling the false discovery rate (FDR)
when performing variable selection.
For more information, see the website below and the accompanying paper: Candes et al.,
"Panning for Gold: Model-X Knockoffs for High-dimensional Controlled Variable Selection",
knockoff 0.3.2 (08/03/2018)
- Fixed bug that caused incorrect knockoff statistics in the presence
of knockoff copies identical to their own original variable.
- Changed scaling of variables for computation of importance measures.
knockoff 0.3.1.1 (06/28/2018)
- Improved algorithm for solving SDP
- Improved algorithm for solving ASDP
- Returning X instead of throwing error in Gaussian knockoffs,
if covariance matrix is not positive-definite
- Minor improvements to package description file
knockoff 0.3.0 (10/17/2017)
- Added support for Model-X knockoffs
- Added importance statistics
- Native support for SDP knockoffs (no need to call Python)
- Model-X knockoffs are used by default
- Cross-validated lasso statistics are used by default
- SDP knockoffs are used by default
- Offset 1 is used by default
- Updated and expanded vignettes
- Add vignette showing how to analyze a real data set (on HIV drug resistance),
including all the preprocessing steps.
knockoff 0.2 (02/04/2015)
- The knockoff procedure is now fully deterministic by default. Randomization
can be enabled if desired.
- Fix numerical precision bug in equicorrelated knockoff creation
knockoff 0.1.1 (12/19/2014)
- Expose the optional 'nlambda' parameter for lasso statistics
- Better documentation for SDP knockoffs
- Minor bug fixes
knockoff 0.1 (12/05/2014)