Package: StableMate
Type: Package
Title: StableMate: stabilized regression via stochastic stepwise variable selection ensemble.
Version: 0.1.0
Author: Yidi Deng
Maintainer: Yidi Deng <yidid@student.unimelb.edu.au>
Description: StableMate (Deng et al.2023) is a regression and variable selection framework in a multi-environment 
    regression setting. Given a response, StableMate distinguish between stable and unstable predictors that have 
    consistent or changing functional dependencies (e.g, fitted regression models) on the response across environments, 
    respectively. Stable predictors entile strong causal implication and can be selected to enforce generalizability of regression models. 
    Unstable predictors (which we refered to as environment-specific predictors in our manuscript) 
    reveals the impact of environment pertubation on the studied system of predictors and response.
    StableMate is built based on the theoretical foundation of Stabilized Regression (Pfister et al. 2021), 
    but implements a different algorithm, Stochastic Stepwise Variable Selection (ST2) (Xin et al. 2012), to discern stability, 
    since the original algorithm is slow and inacurrate when applied to large scale data. We made further modification on the 
    ST2 algorithm to improve computational efficiency, and proposed a new statistical concept called pseudo-predictor that adds on intuitive and 
    accurate benchmark of ST2 selections.
License: None
Encoding: UTF-8
LazyData: true
RoxygenNote: 7.2.3
Imports: 
    Matrix,
    dplyr,
    doParallel,
    ggplot2,
    reshape2,
    ggrepel,
    MASS,
    progress,
    arm
Suggests:
    doMPI,
    gmat,
    igraph,
    knitr,
    rmarkdown
VignetteBuilder: knitr
