When using PROC HPREG with forward stepwise selection, which criterion controls which variables are added to the model at each step?
Select an answer to reveal the explanation.
Short Explanation
Here's the deal — variables enter the model when their F-test p-value falls below SLENTRY=, the significance level for entry. The default is often 0.15 or 0.05. Correlation with residuals, AIC drops, and coefficient size are not the forward-entry rule in PROC HPREG.
Full Explanation
In PROC HPREG forward stepwise selection, the entry criterion is controlled by SLENTRY=, which sets the significance level that a candidate variable's F-test p-value must fall below before the variable can be added to the model; the default is often 0.15 or 0.05. The criterion based on the highest Pearson correlation with the current residuals is wrong because that describes a greedy correlation approach used in matching pursuit algorithms, not PROC HPREG forward selection. The criterion based on reducing AIC below the current model AIC is wrong because that describes AIC-based stepwise selection, which requires a different METHOD= setting. The criterion based on decreasing coefficient magnitude is wrong because coefficient magnitude is not the entry criterion.