Predictor screening differs from response screening. Response screening tests factors one at a time as a predictor of the response. Predictor screening uses bootstrap forest partitioning to evaluate the contribution of predictors on the response. The partition models are built on multiple predictors. Predictor screening can identify predictors that might be weak alone but strong when used in combination with other predictors. See Response Screening topic for details.
Figure 19.1 Example of a Predictor Screening Report
