Import PMML Models

PMML (Predictive Modeling Markup Language) is an XML-based standard for representing data mining results. PMML is designed to enable the sharing and deployment of data mining results between vendor applications and across data management systems. The SAS Model Manager Import PMML Models feature enables users to import PMML models that are produced by using other applications. PMML 4.0 (or later) is supported by SAS Model Manager. Models that are created using PMML 4.0 support DATA step score code. For more information about PMML support in SAS Enterprise Miner, see the SAS Enterprise Miner Help.
Note: Before you import a PMML model, verify that the model type matches the Model Function property setting on the Project Properties panel. For more information, see Specific Properties for a Project.
Note: SAS Model Manager does not support the importing of a PMML file that contains multiple models.
To import a PMML model into SAS Model Manager, follow these steps:
  1. In the Project Tree, navigate to the project's version.
    MMRootthen selectorganizational folderthen selectproject folderthen selectversion folder
  2. Right-click the Models folder and select Import fromthen selectPMML Model File. The PMML Model File window appears.
    PMML Model File Window
  3. Navigate to the location of the PMML file and select the file.
  4. Enter a text value in the PMML model name field. The name can contain letters, spaces, the underscore ( _ ), the hyphen ( - ), and the period ( . ).
  5. Click OK. After the SAS Model Manager processes the model import request, the new model appears in the Models folder of your project's version.
  6. Repeat steps 2 through 5 to import additional PMML files from your folder.
SAS Model Manager generates the score code for the model that is based on the PMML file. The score.sas file is included with the model under the Models node.
Note: After you import a PMML 4.0 model, you can score the model using a scoring task and you can create a model comparison report for the model. You can score tree, regression, and neural PMML models. You cannot define a performance task if the PMML model is set as the champion model.