Skip to main content

Stepwise regression

Stepwise regression is an automated procedure that adds or removes explanatory variables from a multiple regression one at a time, based on a statistical criterion, to search for a parsimonious model among many candidates.

ByHoang TruongUpdated

See it move

Loading infographic...

Predicting rental price from five candidates, forward selection starts empty, then adds floor size first since it alone explains most variation (R² = 0.55). Distance to the city centre enters next, lifting R² to 0.72 with p = 0.01. Bedrooms is tried but only reaches p = 0.35, well above the 0.05 threshold, so it stays out of the final model.

Where it fits
SubjectData Analysis & StatisticsAdvancedTopicMultiple Regression & InterpretationAdvanced