A web-based 'shiny' interface for the 'StepReg' package enables stepwise regression analysis across linear, generalized linear (including logistic, Poisson, Gamma, and negative binomial), and Cox models. It supports forward, backward, bidirectional, and best-subset selection under a range of criteria. The package also supports stepwise regression to multivariate settings, allowing multiple dependent variables to be modeled simultaneously. Users can explore and combine multiple selection strategies and criteria to optimize model selection. For enhanced robustness, the package offers optional randomized forward selection to reduce overfitting, and a data-splitting workflow for more reliable post-selection inference. Additional features include logging and visualization of the selection process, as well as the ability to export results in common formats.
A Shiny application package that provides an interactive web interface for the StepReg package.
StepRegShiny is a companion package to StepReg that provides a comprehensive web-based interface for performing stepwise regression analysis. This package contains the Shiny application components that were previously part of the main StepReg package, allowing for better separation of concerns and independent development of the web interface.
# Install from CRAN (when available)
pak::install_package("StepRegShiny")
install.packages("StepRegShiny")
# Or install from GitHub
devtools::install_github("JunhuiLi1017/StepRegShiny")
library(StepRegShiny)
# Launch the Shiny application
StepRegGUI()
This package requires the StepReg package to function properly. Make sure you have StepReg installed.
MIT + file LICENSE