Configuring Model Manager Java Services Options :: SAS(R) Decision Manager 3.1: Administrator's Guide

Overview of Configuring Model Manager Java Services Options

The Model Manager Java Services Options setting in SAS Management Console enables you to modify model management configurations. The configurations can be modified for reporting, for metadata tables usage when publishing a scoring function, for SAS code debugging, SAS system options, In-database connection and publishing options, and performance options for the SAS High-Performance Analytics procedures.

To modify the settings for the report options:

Log on to SAS Management Console as a SAS administrator.
On the Plug-ins tab, navigate to Application ManagementConfiguration ManagerSAS Application InfrastructureEnterprise Decision Manager 3.1.
Right-click Model Manager JavaSvcs 14.1 and select Properties.
Click the Settings tab and then select Model Manager Java Services.
Select one of the following options to view and configure the available settings.
Click OK.
For changes to take effect, you must restart the web application server.

Report Options

The Report Options setting in SAS Management Console enables you to modify the SAS Decision Manager configurations for the dashboard reports, for model retrain reports, and for performance monitoring. These reports are available on the Reports page for a project in SAS Decision Manager.

To modify the report options setting:

Specify the formats that are available when a user creates model retrain reports. The default values are RTF, PDF, and HTML. You can remove any of the default values so that they are not available in SAS Decision Manager.
Specify the report styles that are available when a user creates the model retrain reports and dashboard reports. You can add or remove SAS styles. The default values are SAS default, Seaside, Meadow, and Harvest. For more information about SAS styles, see Style Templates in SAS Output Delivery System: User’s Guide.
Specify a value for a random seed to be used by performance definitions for models that have an interval target. The default value is 12345. The random seed is the initial seed for the random number generator used for sampling the input data set.
Specify a value for the sample size that is used by performance definitions for models that have an interval target. The default value is 1000. The sample size for models with an interval target is the number of observations from the input data set.
Specify a value for the sample size that is used by performance definitions for characteristic and stability analysis. The default value is 10000. The sample size for characteristic and stability analysis is the number of observations from the input data set.
Select Yes or No to specify whether to use the temporary tables on the High-Performance Analytics appliance for performance monitoring. The default is Yes.
Specify a value for the correlation coefficient (rho) that is used by the binomial test to estimate the probability of default (PD) levels. The default value is 0.04.

Publish Scoring Options

The Publish Scoring Options setting enables you to indicate that the metadata tables be populated in the target database when publishing a scoring function. The default is Yes. During the installation and configuration process of the database, the metadata tables must be created in the database if this setting is set to Yes. If you plan to use only the SAS Embedded Process publish method to publish scoring model files, this setting can be ignored, and you do not need to create the metadata tables during the database configuration process.

For information about the database configurations, see Preparing a Data Management System for Use with SAS Model Manager in SAS In-Database Products: Administrator’s Guide. For more information about publishing models to a database, see Publishing Models to a Database or Hadoop in SAS Decision Manager: User’s Guide.

Debug Options

The Debug Options setting enables you to use the debug options when executing SAS code within SAS Decision Manager. The default value is No.

However, the Debug Options setting does not work for scoring tests, performance definitions, and model retrain definitions. To enable debug options with scoring tests, you must add the following line of code to the appserver_autoexec_usermods.cfg file in the \sasconfigdir\Lev#\SASApp\WorkspaceServer\ directory:

options mprint symbolgen notes;

Note: You can also add this code using the Edit Start-up Code feature in SAS Decision Manager.

Valid Variable Name Options

The VALIDVARNAME= system option specifies the rules for valid SAS variable names that can be created and processed during a SAS session. The Valid Variable Name Options setting enables you to set the VALIDVARNAME system option to ANY when executing SAS code. To do this, you select Yes. The default value is No.

However, the Valid Variable Name Options setting does not work for scoring tests, performance monitoring, or retraining jobs. To use the VALIDVARNAME system option with these features, you must add the following line of code to the sasv9_usermods.cfg file in the \sasconfigdir\Lev#\SASApp\WorkspaceServer\ directory:

options validvarname='any';

Note: You can also add this code using the Edit Start-up Code feature in SAS Decision Manager. For more information, see Edit Start-Up Code in SAS Decision Manager: User’s Guide.

When the VALIDVARNAME system option is set to V7, variable names with letters of the Latin alphabet, numerals, or underscores are valid. If you use any other characters, then you must express the variable name as a name literal, and you must also set VALIDVARNAME=ANY. If the name includes either the percent sign (%) or the ampersand (&), then you must use single quotation marks in the name literal in order to avoid interaction with the SAS Macro Facility. For more information, see VALIDVARNAME= System Option in SAS System Options: Reference.

Note: You must have set the VALIDVARNAME system option to ANY in order to use PMML models that were created using SAS Enterprise Miner in scoring tests, performance monitoring, and reporting within SAS Decision Manager.

In-Database Options

The In-Database Options settings enables you to specify the publish type, database connection settings, and publish settings that are used when publishing models to a database using SAS Decision Manager.

To modify the settings for the in-database options:

Select a method to publish models to the database for scoring. The default publish type is the SAS Embedded Process publish method.
Select a database type.

Specify values for database settings that are required to publish to the selected database type.

Here are the available database settings according to the publish method and database type:

Database Settings	SAS Embedded Process	Scoring Function
Server name	Teradata Oracle Netezza Greenplum DB2 Hadoop SAP HANA	Teradata Netezza Greenplum DB2
HDFS directory path	Hadoop	Not applicable
Database name or instance number	Teradata Oracle Netezza Greenplum DB2 SAP HANA	Teradata Netezza Greenplum DB2
User ID	Teradata Oracle Netezza Greenplum DB2 Hadoop SAP HANA	Teradata Netezza Greenplum DB2
Server user ID	Not applicable	DB2
Schema	Oracle Greenplum DB2 SAP HANA	Greenplum DB2

Specify to use the model manager table when publishing. The default value is No. When you are publishing the scoring model files to a database using the SAS Embedded Process publish method, the files are by default stored in the table sas_model_table. If the Use model manager table value is set to Yes, the scoring model files are stored in the table sas_mdlmgr_ep. These tables are located in the target database. This setting enables users to separate the SAS Decision Manager scoring model files from the SAS model scoring files when using the SAS Embedded Process publish method.

Note: The sas_mdlmgr_ep table must be created before setting this option to Yes. For more information, see Configuring a Database in SAS In-Database Products: Administrator’s Guide.
Specify to force the republish of model scoring files by default when using the SAS Embedded Process publish type. The default value is No. If you set this setting to Yes, then the Replace scoring files that have the same publish name option in the Publish Models window in SAS Decision Manager is selected by default.
Select the default format of the model publish name when using the SAS Embedded Process publish method. The format selected determines the value that appears for the publish name in the Publish Models window in SAS Decision Manager. The scoring function publish method publish name defaults to the model name.
Specify a directory path to store the temporary scoring files. If a value is not specified, the SAS work directory is used by default. However, if the directory is not specified and you select the Display detailed log messages option when publishing to a database the SASUSER directory is used.

For information about the database configurations, see Preparing a Data Management System for Use with SAS Model Manager in SAS In-Database Products: Administrator’s Guide. For more information about publishing models to a database, see Publishing Models to a Database or Hadoop in SAS Decision Manager: User’s Guide.

Performance Options

The Performance Options setting contains the performance parameters for the PERFORMANCE statement to use the SAS High-Performance Analytics procedures. Currently only the Teradata and Greenplum database types support SAS High-Performance Analytics.

The PERFORMANCE statement defines performance parameters for multithreaded and distributed computing, passes variables about the distributed computing environment, and requests detailed results about the performance characteristics of a high-performance analytics procedure.

The following performance options can be specified for the PERFORMANCE statement.

Option	Description	Default Value
Commit	Specifies an integer to request that the High-Performance Analytics procedure writes periodic updates to the SAS log.	10000
CPU count	Specifies how many processors the procedure should assume are available on each host in the computing environment. You can enter the value of ACTUAL, or enter an integer between 1 and 256.	ACTUAL
Database server	Specifies the name of the server for the database as defined through the hosts file and as used in the LIBNAME statement.
Details	Requests a table that shows a timing breakdown of the procedure steps.	No
Timeout	Specifies the time-out in seconds for a High-Performance Analytics procedure to wait for a connection to the appliance and establish a connection back to the client.	120
Host name	Specifies the name of the appliance. If a value for the Host option is specified, it overrides the value of the grid host environment variable.
Installation directory	Specifies the name of the directory in which the High-Performance Analytics shared libraries are installed on the appliance.
Install location	Specifies the name of the directory in which the High-Performance Analytics shared libraries are installed on the appliance. If a value is specified for the Installation directory option, it overrides this option.
Number of nodes	Specifies the number of nodes in the distributed computing environment, provided that the data is not processed alongside the database. You can enter an integer or you can specify a value of ALL if you want to use all available nodes on the appliance without oversubscribing the system.
Number of threads	Specifies the number of threads for analytic computations. This option overrides the SAS system option THREADS \| NOTHREADS. If you do not specify a value for this option, the number of threads are determined based on the number of CPUs on the host on which the analytic computations execute.
Grid host	Specifies the host name for the grid. If a value for the Host option is specified, it overrides the value of the grid host environment variable.
Grid RSH command	Specifies the command to run a remote shell.
Grid reply host	The host name of the client node to which the grid connects.
Grid port range	Specifies the range of parts that are permitted by the firewall.
Grid path	Specifies the local directory path for the grid node.
Grid mode	Specifies whether the HPFORECAST procedure runs in symmetric (SYM) mode or asymmetric (ASYM) mode. The default is symmetric (GRIDMODE=SYM).	Symmetric

For more information about High-Performance Analytics, see SAS High-Performance Analytics Server: User's Guide.

Factory Miner Options

If SAS Factory Miner and SAS Decision Manager are licensed at your site, you can register projects segments and models from SAS Factory Miner to the SAS Model Manager model repository. The project segments and models are created within a portfolio. Use the Automatically set champion setting to automatically set the project champion in the model repository when a model is set as the champion for a project segment in SAS Factory Miner. The default value is No. For information about integrating SAS Factory Miner with SAS Decision Manager, see SAS Factory Miner: Administration and Configuration.