Predicted Values with Backward Selection

[Shivi82]

Dear All,

i am on one of the courses offered from the SAS institute for learning & taking some clues which i will further add to one of my projects. this should be an easy fix i am guessing.

KIndly see the code that i have built:

proc reg data = statdata.bodyfat2;
model PctBodyFat2= age weight height neck chest abdomen hip thigh knee ankle biceps forearm wrist
/selection=backward;
id weight;
run;

As we all know the the elimination method will throw insignificant var one by one. My question here is that along with the elimination method i need to add "p" that calculates predicted values for a given value of x.

Hence the code should look like

proc reg data = statdata.bodyfat2;
model PctBodyFat2= age weight height neck chest abdomen hip thigh knee ankle biceps forearm wrist/p /selection=backward;
id weight;
run;

However this throws an error. Kindly let me know if this is doable.

Regards, Shivi

[Ksharp · Accepted solution]

Check Documentation , Found this :

proc reg data = statdata.bodyfat2;
model PctBodyFat2= age weight height neck chest abdomen hip thigh knee ankle biceps forearm wrist
/selection=backward;
id weight;

output out=b p=yhat ;
run;

[Shivi82]

Thank you this was the solution i was searching for.

Regards, Shivi