This can be done based on various statistics by using the SCORE statement in PROC LOGISTIC as described and illustrated in this note. See in particular the use of the FITSTAT option in the SCORE statement which provides a table of various statistics. The simple Error Rate for the two data sets compares the proportions of incorrectly classified observations in each data set. For details on the statistics, see "Model Fitting Information" in the Details section of the LOGISTIC documentation.
... View more