What's New in SAS Model Manager 12.1

Overview

SAS Model Manager 12.1 has the following new features and enhancements:

enhanced support to manage documents
ability to use tables from a local or network drive
ability to import SAS/STAT linear models and SAS High-Performance Analytics models
ability to create DATA step score code for PMML models
ability to schedule scoring tasks
enhanced support for champion and challenger models
enhanced support for publishing models
enhanced performance monitoring and reporting
ability to create aggregated reports
ability to create Basel model validation reports
ability to view dashboard reports and manage dashboard report definitions
ability to perform model management tasks using SAS Model Manager Workflow Console
ability to associate milestones with workflow activities

Enhanced Support to Manage Documents

SAS Model Manager provides the ability for the Documents folder to be created at the organizational folder level and project folder level, in addition to the Documents folder that is created by default when a new version is created. You can also create subfolders within the Documents folder. You can associate documents with a modeling project by attaching them to the Documents folder or a subfolder.

Ability to Use Tables from a Local or Network Drive

SAS Model Manager now provides the ability to edit the SAS Model Manager start-up code so that you can submit a LIBNAME statement to define a libref or other SAS code. Therefore, you can use tables from a local or network drive instead of data sources that are registered in the SAS Metadata Repository.

Ability to Import SAS/STAT Linear Models and SAS High-Performance Analytics Models

SAS Model Manager now supports importing SAS/STAT linear models and SAS High-Performance Analytics models from a SAS package file (.SPK). Models that have a DATA step score code type can also be scored, published, and included in performance monitoring.

Ability to Create DATA Step Score Code for PMML Models

SAS Model Manager now supports the creation of DATA step score code for PMML models, which enables a PMML model to be included in scoring tasks, reporting, and performance monitoring. PMML 4.0 (or higher) is supported by SAS Model Manager for creating DATA step score codes when importing models.

Ability to Schedule Scoring Tasks

SAS Model Manager uses the scoring task scheduler to enable a user to schedule a scoring task to run at a certain date and time on available servers, and to specify where to save the scoring task output. You can then view the job history for a scoring task.

Enhanced Support for Champion and Challenger Models

SAS Model Manager now provides support for challenger models. When a model is set as the project champion, the version that contains the champion model is automatically set as the default version. Models can then be selected as challengers to the project champion model. Only one champion model can be set per project. You can monitor and publish the challenger models in addition to the champion model.

Enhanced Support for Publishing Models

SAS Model Manager now supports the selection of multiple models when publishing to the SAS Metadata Repository and to a database. The SAS Embedded Process publish method can be used to publish model scoring files to a database. The databases that are currently supported for publishing models using the SAS Embedded Process are Teradata, Greenplum, Oracle, and DB2.

Enhanced Performance Monitoring and Reporting

The following enhancements have been made to performance monitoring and reporting:

Performance monitoring can now be performed for projects that are associated with a prediction model function that has an interval target.
Performance monitoring and scoring can now be performed on an appliance that is configured to use High-Performance Analytics. Teradata and Greenplum currently support the High-Performance Analytics configuration.
If the performance data source that you specified when you created a performance task contains data in the scoring output variables, you can choose to omit the running of the scoring task when executing a performance task.
The Interval Target Variable report has been added to the available model comparison reports. The Interval Target Variable report creates two plots for you to view the actual versus predicted values for a model and the actual versus residual values for a model. The Interval Target Variable report can be created only for prediction models.
The Training Summary Data Set report has been added to the training summary reports. The Training Summary Data Set report creates frequency and distribution charts to validate data set variables.
The Champion and Challenger Performance Report has been added to the available reports. If you want to determine how a challenger model performs in comparison to the champion model, you can define and execute performance monitoring reports for a champion model and a challenger model, and then create a Champion and Challenger Performance report to compare the two models.

Ability to Create an Aggregated Report

SAS Model Manager now enables you to create an aggregated report to combine multiple reports from the Reports folder into a single report. The format of the aggregated report can be PDF, HTML, or RTF. Aggregated reports are stored in the Documents folder.

Ability to Create Basel Model Validation Reports

Basel reports have been added to the available reports to provide several statistical measures to validate the stability, performance, and calibration of models. Model validation reports can be created for the two key types of Basel II risk models: probability of default (PD) and loss given default (LGD).

Ability to View Dashboard Reports and Manage Dashboard Report Definitions

SAS Model Manager provides the ability to launch a browser to view the Dashboard Reports from the SAS Model Manager client and the SAS Model Manager Workflow Console. You can also create, edit, and delete project dashboard report definitions from the SAS Model Manager client.

Ability to Perform Model Management Tasks Using SAS Model Manager Workflow Console

SAS Model Manager now enables you to integrate SAS Workflow with some of the model management tasks that are normally performed in the SAS Model Manager client. Workflow process definitions can be configured to use model management components with the workflow activities.

Here are the model management tasks that can be performed:

import models
view models
set a project champion model and flag challenger models
publish models
add, view, or delete attachments
view reports
view performance

Ability to Associate Milestones with Workflow Activities

SAS Model Manager now enables you to associate milestones with workflow activities as part of the workflow process definition. Activities that are associated with a milestone appear on the Workflow Milestones tab in the version details section of the SAS Model Manager client. The Workflow Milestone report has also been added to the available reports.