The SCORE statement creates a new SAS data set containing predicted values and optionally residuals for data in a new data set that you name. If you do not specify a DATA= data set, then the input data are scored. If you have multiple data sets to predict, you can specify multiple SCORE statements. If you want to create a SAS data set in a permanent library, you must specify a two-level name. For more information about permanent libraries and SAS data sets, see SAS Language Reference: Concepts.
When a BY statement is used, the score data set must either contain all the BY variables sorted in the order of the BY variables or contain none of the BY variables. If the score data set contains all of the BY variables, then the model selected for a given BY group is used to score just the matching observations in the score data set. If the score data set contains none of the BY variables, then the entire score data set is scored for each BY group.
All observations in the score data set are retained in the output data set. However, only those observations that contain nonmissing values for all the continuous regressors in the selected model and whose levels of the classification variables appearing in effects of the selected model are represented in the corresponding classification variables in the procedure’s input data set are scored. All the variables in the input data set are included in the output data set, along with variables containing predicted values and optionally residuals.
Details on the specifications in the SCORE statement follow: