A data scientist uses PROC HPREG with SELECTION=STEPWISE and SLENTR…

Question

A data scientist uses PROC HPREG with SELECTION=STEPWISE and SLENTRY=0.05 SLSTAY=0.10. What do these two significance level options control?

Accepted Answer

SLENTRY=0.05 is the p-value threshold for a variable to enter the model; SLSTAY=0.10 is the p-value threshold for a variable to remain in the model — Here's the deal — a is correct because in PROC HPREG (and PROC REG) stepwise selection: SLENTRY= specifies the significance level (p-value) that a candidate variable's F-test must achieve to enter the model, and SLSTAY= specifies the significance level that an included variable must maintain to remain in the model. Using SLENTRY < SLSTAY (0.05 < 0.10) ensures the entry criterion is stricter than the stay criterion, preventing cycling.

Answer

SLENTRY=0.05 is the minimum R-squared increase required to add a variable; SLSTAY=0.10 is the minimum R-squared to keep a variable

Answer

Both SLENTRY and SLSTAY are redundant — stepwise uses only the smaller value

Answer

SLENTRY=0.05 limits the number of variables entering to 5%; SLSTAY=0.10 keeps 10% of all variables

A data scientist uses PROC HPREG with SELECTION=STEPWISE and SLENTRY=0.05 SLSTAY=0.10. What do these two significance level options control?

Short Explanation and Infographic

Full Explanation