All the predictive modeling
nodes enable you to score the training, validation, test, and scoring
data sets in conjunction with training. To score other data sets,
especially new data not available at the time of training, use the
Score node.
Each predictive modeling
node generates SAS DATA step code for computing predicted values.
The Score node accumulates the code generated by each modeling node
that precedes the Score node in the flow diagram. The Score node then
packages all the scoring code into a DATA step that can be executed
to score new data sets. The scoring code can be saved for use in the
SAS System outside of SAS Enterprise Miner.
The Score node also
handles:
-
code for transformations generated
by the Transform Variables node
-
code for missing-value imputation
generated by the Impute node
-
code for cluster assignment generated
by the Cluster node
-
code for decision processing.
You can use a SAS Code
node following the Score node to do additional processing of the scored
data. For example, if you used the Model Comparison node to choose
a decision threshold, you could apply the threshold in a SAS Code
node.