Forward Selection

A model building technique where a model is built by starting with no predictors in the model. Variables are then added to the model one at a time if found to be significant contributors to model quality (accuracy). The most significant variable is added at each step and the model is recalculated at each step.