I have a training dataset with (x1,x2,x3,y) and these contain some missing observations. I ran proc logistic on this training data and got the parameter estimates (bo, b1, b2, b3).

I tried to apply these estimates(bo, b1, b2, b3) on the test dataset which does not contain any missing observation using proc logistic inmodel the results show lot of missing data and this is the message from the log " Work.Test data set are not scored because they have class levels that are missing or are not present in the analysis data set"

Any suggestion how to score on test dataset without producing missing data even when training dataset contains missing data ? Thanks in advance.


Check your scoring input data for character variables which have different classes than in training data. I think that for numeric variables it is no problem but for char variables which are factored using certain codes this can be problem. SAS treats missing data differently if variable is char or num.

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.