Is It possible to delete outlier observation under rstudent condition in reg procedure ?

[LeeSeongWoo_]

Hello . Thank you for your time. 

 I have some terrible question. 

   I have to clear the observation that out of (-2,2) rstudent . 

Does anybody know how to delete those observation out of (-2,2) ,by coding ? there are too many to find and delete manually...

[ballardw]

From you graph it appears that you may have a variable rstudent with values of concern. If that is the case then what you want is fairly simple:

Data want;

    set have;

    where -2 le rstudent le 2;

run;

[LeeSeongWoo_]

Thank you for your reply . but This graph is part of the proc reg result . 

for example :

  proc reg data = work.aa; 

  model y=x1 x2 x3; 

 run; 

so i don't have rstudent variable . 

what i want to know is how can i make these variable ?

Thank you for reading 

[Ksharp · Accepted solution]

Use OUTPUT statement to save rstudent:

proc reg data=sashelp.class;
model weight=age height;
output out=want rstudent=rstudent cookd=cookd;
quit;

and proc sort it ,delete obs you don't want.

[ballardw]

You can use dataset options on output data sets in procedures that create data sets directly:

proc reg data=sashelp.class;
model weight=age height;
output out=want (where=( 2 le rstudent le 2))
      rstudent=rstudent cookd=cookd;
quit;