NEWS.md

In flevr: Flexible, Ensemble-Based Variable Selection with Potentially Missing Data

v0.0.4

Major changes

• Bugfixes for xgboost objects and extract_importance_SL
• pkgdown site

Minor changes

• Add @examples and @return sections in documentation for all functions
• Use inherit to avoid redundant documentation of parameters/sections

v0.0.3

Major changes

• Added importance extractor for polymars objects

Minor changes

• Updated internal checking to inherits

v0.0.2

Major changes

• Updated PFP and FDR control for intrinsic selection
• Added functions spvim_vcov (compute variance-covariance matrix for SPVIMs for intrinsic selection); pool_spvims (pool point estimates and obtain Rubin's rules-based variance estimates for multiple imputation-based procedures).
• Added a dataset, biomarkers, for use in vignettes

v0.0.1

Major changes

• Added function extract_importance_svm, which computes extrinsic importance from a ksvm object
• Added a vignette on basic usage of the package

Minor changes

None

v0.0.0.9000

Major changes

The first version of the flevr package! This package provides functions for variable selection using flexible ensembles. These functions can be used on both complete-case and multiply-imputed data. The two procedures considered are based on extrinsic variable importance (estimated using the Super Learner) and intrinsic variable importance (estimated using the Shapley Population Variable Importance Measure). Both can also be embedded within stability selection using stabs.

Minor changes

None

flevr documentation built on June 22, 2024, 7:33 p.m.