The GAM Procedure
- Overview
- Getting Started
-
Syntax
-
DetailsMissing ValuesNonparametric RegressionAdditive Models and Generalized Additive ModelsForms of Additive ModelsEstimates from PROC GAMBackfitting and Local Scoring AlgorithmsSmoothersSelection of Smoothing ParametersConfidence Intervals for SmoothersDistribution Family and Canonical LinkDispersion ParameterComputational ResourcesODS Table NamesODS Graphics
-
Examples
- References
The SCORE statement calculates predicted values for a new data set. All the variables in the DATA= data set are included in the OUT= data set, along with the predicted values. The predicted values consist of predicted responses after the inverse link function transformation, predicted values of all smoothing terms, and predicted values on the link scale. Predicted values are computed for observations with missing response values whose values of the specified explanatory variables are nonmissing, and whose values of the specified smoothing variables are within the smoothing ranges of the fitted model. The predicted variables use the same naming convention as the OUTPUT statement. If you have multiple data sets to score, you can specify multiple SCORE statements.
The following options must be specified in the SCORE statement: