Overview of Publishing Models

SAS Model Manager provides a comprehensive publishing environment for model delivery that supports sharing performance and scoring data. SAS Model Manager publishes models to different channels, and to the SAS Metadata Repository. SAS Model Manager can also publish classification, prediction, and segmentation (cluster) models that have the score code type of DATA step to a database or Hadoop. Models that have a score code type of analytic store can be published only to Hadoop and Teradata. Application software, such as SAS Data Integration Studio or SAS Enterprise Guide, enables you to access models through the SAS Metadata Server and to submit on-demand and batch scoring jobs.
SAS Model Manager publishes models to defined publication channels. Authorized users who subscribe to a channel can choose to receive email notifications when updated models are ready to deploy to testing or production scoring servers, and are published to a publication channel. From a publication channel, you can extract and validate the scoring logic, deploy champion models to a production environment, and monitor the performance of your models.
Models can also be published from the Models page. You can publish champion and challenger models from a model project to the SAS Metadata Repository. The publish history of models can be viewed on the Models page and on the Published tab on the History page. You can also remove models that have been published to a database.
Last updated: June 12, 2017