PROC GLM Error Message

[dhrumil_patel]

I am receiving the following error in the log after submitting this PROC GLM code.

proc glm data=dsn;
by segment;
class trim model category subsegment;
model price = trim model age category subsegment;
quit;

The error is:

"One or more variables are missing or freq or weight is zero on every observation."

I know the error is coming from the subsegment variable because I added the variables one at a time before getting an error with subsegment. 

I'm not sure how to diagnose the source of the error. Subsegment does have a 2% missing rate, but not all values are missing. Besides, I thought PROC GLM would simply exclude observations with missing input values. 

Can anyone provide help on the source of this error?

Thank you.

Dhrumil Patel

[Reeza]

Post your log and look at output from proc freq on variable. My guess is you misspelled the variable name?

[dhrumil_patel]

Thanks for your reply.

I checked it wasn't from misspelling the variable name. It was because of my BY statement. There were combinations of segment and subsegment where the values were completely missing.

Dhrumil

[PGStats · Accepted solution]

GLM does separate analysis for every BY-value. Check that subsegment is not entirely missing for one of the segments.