PROC PSCORE and PMML Support

Overview

PMML is an XML markup language that was developed to exchange predictive and statistical models between modeling systems and scoring platforms. Users can import the majority of standard-compliant PMML models and score them within a SAS environment via the SAS PSCORE procedure.

PROC PSCORE Functionality

The SAS PSCORE procedure generates SAS DATA step score code that is functionally equivalent to the PMML model. The generated score code can be executed on all platforms that are supported by SAS to score the data sets. You can submit the score code in SAS Enterprise Miner via the Program Editor, SAS Enterprise Miner Project code, or within a SAS Enterprise Miner Process Flow Diagram, via the SAS Code node. However, the SAS Enterprise Miner UI environment is not necessary to run the score code.

Supported Versions

PROC PSCORE currently supports the use of PMML 4.2. Earlier versions of PMML are not supported for use with PROC PSCORE.

Supported PMML Models

PROC PSCORE supports the following types of PMML models:
  • Regression
  • Trees
  • Neural Networks
  • Clustering models
  • Scorecard
  • Vector Machine
  • Naïve Bayes
  • Baseline models
The following models are supported on an experimental basis:
  • Time Series
  • General Regression

Requirements for PROC PSCORE

In order to use PROC PSCORE, you must have SAS 9 or later, a well formed PMML modeling file, and Write access to the output directory for the DATA step score file. A SAS Enterprise Miner license is not necessary to run PROC PSCORE.

PROC PSCORE Usage

PROC PSCORE PMML FILE = “<full-pathname-of-PMML-file>"
     DS FILE = "<full-pathname-of-output-DS-file>"

PROC PSCORE Example

/*Run the PSCORE procedure on a generated PMML file*/
PROC PSCORE PMML FILE = "C:\temp\heart_pmml1.xml"
     DS FILE = "C:\temp\ds_heart_score.sas";
  run;
Last updated: June 12, 2017