To create the monitoring
reports, you specify a performance definition to generate SAS code.
You then execute the generated code or create a schedule to execute
the generated code on a specific day and time. Execution of the generated
code creates the SAS data sets that are used to display reports: either
the monitoring reports from the version Performance page,
or the Monitoring report or Champion and Challenger Performance report
that you create from the New Report window.
To edit the performance
definition:
-
Click
Edit
Definition and select a champion or challenger model.
Click
Next.
-
Select a SAS Application
Server.
-
Select one or more output
variables for stability analysis. To select all output variables,
click
All.
-
Select one or more input
variables for characteristic analysis. To select all input variables,
click
All. Click
Next.
-
Choose the data processing
method:
-
To run a standard environment,
select Standard configuration.
To run the score code
in the performance monitor job, select the Run model score
code check box.
-
To run the performance monitoring
definition in a High-Performance Analytics environment, select High-performance
configuration.
Note: The score code is not run
when High-performance configuration is selected.
-
Decide to use either
the static or dynamic data sources, and then specify the data source
information.
Note: Ensure that the data source
information is complete before saving the definition. If you start
adding information for static data sources and then decide to use
dynamic data sources instead, be sure to delete the information added
for static data sources before adding the dynamic data source information,
and vice versa.
To use static data
sources:
-
Click
.
Note: If you are adding multiple
tables in the first performance definition, the first table that you
select is the baseline performance data table.
-
Click the empty cell
in the
Data Source column.
-
Click
Browse and
select a performance data source. Click
OK.
-
Click the empty cell
in the
Collection Date column and click
. Select a date. The date can be any date in the time
period when the performance data was collected.
-
To add a label for the
date, enter the label name in the
Report Label column.
The report label represents the time point of the performance data
source. Because the report label appears in the performance charts,
use a label that has not been used for another time period, is short,
and is understandable (for example, Q1).
Note: Duplicate report labels result
in previous performance results being overwritten.
-
(Optional) Select a
data source and click
to verify that the selected input variables and target
variable are included in the performance data source.
-
(Optional) Repeat the
above steps to add multiple performance data sources to the performance
definition.
-
(Optional) To delete
a data source from the performance definition, select the data source
and click
.
To use dynamic data
sources:
-
Click
to select a data source library.
-
(Optional) Specify the
prefix to remove from the data source names in the selected library.
The data source name is used for the report label. You can remove
the prefix so that it does not show as part of a report label on the
charts.
-
(Optional) Select
Generate
dashboard reports after the performance monitoring has completed.
The dashboard definition must already exist for this option to work.
-
-
(Optional) Either specify
values for the alert and warning conditions or accept the defaults.
-
(Optional) To send the
results by e-mail, click
. A new row is added to the table.
-
-
Select either
Yes or
No if
you want an alert or warning to be sent by e-mail when alert or warning
thresholds have been exceeded.
-
Select either
Yes or
No if
you want a completion notice with the job status to be sent by e-mail
every time the report runs.
-
To execute a performance
definition:
-
Select the
Performance page
for the project.
-
Click
.
-
After the performance
monitoring has been completed, a confirmation message appears. Click
Close.
-
Click the
Results tab
to view the performance results.
Note: You can check the status
of a job by clicking
and then selecting the
Results tab
or the
Job History tab.
Note: You can overwrite or delete
previously created performance data sets.