StepReg

>> An R package designed to streamline stepwise regression analysis while promoting best practices

StepReg R shiny

Description: Stepwise regression is commonly used for model selection, but the lack of a unified tool supporting various regression types, selection strategies, and metrics complicates its proper use. This study introduces StepReg, an R package designed to streamline stepwise regression analysis while promoting best practices. StepReg supports multiple regression types, integrates popular selection methods, and includes key selection metrics. Additionally, StepReg allows users to select multiple strategies and metrics for efficient model selection, visualize variable selection, and export results in various formats. However, StepReg should not be used for statistical inference unless the variable selection process is explicitly accounted for, as this can invalidate the results. This issue does not arise when StepReg is used for prediction. StepReg was validated using public datasets in SAS to ensure accuracy and features an interactive Shiny application to enhance its accessibility.