fastFGEE - Fast Functional Generalized Estimating Equations via a One-Step Estimator

Fits functional generalized estimating equations for longitudinal functional outcomes and covariates using a one-step estimator that is fast even for large cluster sizes or large numbers of clusters. The package supports quasi-likelihoods derived from a range of families, common link functions, and several working correlation structures. Uncertainty quantification is based on sandwich variance estimators and bootstrap procedures that remain valid even when the working correlation is incorrectly specified. The package provides an implementation of the method described in Loewinger et al. (2025) <https://pmc.ncbi.nlm.nih.gov/articles/PMC12306803/>. For irregularly spaced AR(1) precision matrices, the package can optionally use the archived package 'irregulAR1'; if needed, it can be obtained from the CRAN Archive at <https://cran.r-project.org/src/contrib/Archive/irregulAR1/>.

Last updated

4.60 score 503 downloads

studyStrap - Study Strap and Multi-Study Learning Algorithms

Implements multi-study learning algorithms such as merging, the study-specific ensemble (trained-on-observed-studies ensemble) the study strap, the covariate-matched study strap, covariate-profile similarity weighting, and stacking weights. Embedded within the 'caret' framework, this package allows for a wide range of single-study learners (e.g., neural networks, lasso, random forests). The package offers over 20 default similarity measures and allows for specification of custom similarity measures for covariate-profile similarity weighting and an accept/reject step. This implements methods described in Loewinger, Kishida, Patil, and Parmigiani. (2019) <doi:10.1101/856385>.

Last updated

2.00 score 2 scripts 197 downloads

sMTL - Sparse Multi-Task Learning

Implements L0-constrained Multi-Task Learning and domain generalization algorithms. The algorithms are coded in Julia allowing for fast implementations of the coordinate descent and local combinatorial search algorithms. For more details, see a preprint of the paper: Loewinger et al., (2022) <arXiv:2212.08697>.

Last updated

1.00 score 8 scripts 195 downloads