In modeling process, you need to impute missing values in the validation data set the same way you did in training sample. In SAS, there is an easier way to accomplish this task using PROC STDIZE.
SAS Code
proc stdize data=training out=abc method=median reponly outstat=info;
var v1 v2 v3;
run;
proc stdize data=validation out=valid_updated reponly method=in(info);
var v1 v2 v3;
run;
Post a Comment