Pipelines consist of
a series of nodes that are run in a sequence to produce forecasts
based on the models and settings that you choose for the project.
After you first create a project, the pipeline is set up in the Pipelines tab.
The pipeline is labeled By default, the initial pipeline for a project uses the template specified when the project was created. You can create new pipelines using different templates and you can make changes to the initial pipeline. |
|
The pipeline cannot be run until you assign the required variable roles as described in Assigning Variable Roles.
After running the pipeline, you can use the Model Comparison node to determine the best performing node, or champion. The results for each pipeline are derived from the champion modeling node.
As you look at a pipeline, hold your pointer over each node to see its status.
Initialized. Successful. Pending is waiting for
other nodes in the pipeline to complete before starting.Running.
You cannot make any changes to the node until it finishes running.Failed status. Out-of-date status,
and the pipeline must be run again.Canceled. This can occur for one of the following
reasons:
After a pipeline is run, the node that contains
the champion model within the pipeline is indicated by the icon.
The node that contains the champion model for the
project is indicated by the icon. Sometimes this icon does not show when the node image is too small.
In this case, you need to enlarge the pipeline image with the mouse scroll wheel.
If you place an Interactive Modeling node after any modeling node in a pipeline, the Interactive Modeling node shows the champion model if either of the following are true:
Right-click any node in a pipeline to open a menu of actions.
Successful are
run first. After the preceding nodes complete successfully, the selected
node runs.If you rename a modeling node that is connected to an Ensemble node, the Ensemble node becomes out-of-date. The Ensemble node and the rest of the pipeline need to be run again.
Right-click the Data node to open these actions in the menu.
Right-click any modeling node to get these actions in the menu.
Adds an Interactive Modeling node or Ensemble node after the modeling node. These options are in the Postprocessing sub menu. You can also add a Save Data node from the Miscellaneous sub menu.
Removes the node from the pipeline. If it is the only modeling node in the pipeline, then the Model Comparison and Output nodes are also removed.
Adds the modeling node to your clipboard. You can then paste the modeling node to any pipeline in your project. Any changes that you have made to the modeling node are also included when you paste the modeling node.
Pastes an Interactive Modeling node, Ensemble node, or Save Data node that you have copied into your clipboard.
Opens the Forecast Viewer, which provides a plot of the time seriesan aggregation of transactional data into specified time intervals and sorted according to unique combinations of the default attributes (BY variables), including historical and forecasta numerical prediction of a future value for a specified time period for each unique combination of BY variable values data for the project. For more information, see Opening the Time Series and Forecast Viewers.
Opens the code editor for the node. This option is available only for pluggable modeling nodes.
Changes the settings for eventsan incident that disrupts the normal flow of any process that generates the time series. Examples of events are holidays, retail promotions, and natural disasters. for the node in this pipeline. See Changing Event Usage in Modeling Nodes for more information.
Saves the node, with any settings you have made in the pane on the right, to The Exchange. This makes the saved node available for other projects.
Opens the results of the modeling node, including an execution summary, MAPE distribution, and other relevant data. For more information, see Viewing the Results for a Modeling Node.
Downloads a PDF of the results for a modeling node after it has completed successfully. For more information, see Generating Output from SAS Model Studio in SAS Visual Forecasting: Overview.
After the modeling node has run, this action opens a read-only view of the code for the modeling node .
For segmented pipelines, you have these nodes to work with.
Merges the results produced by the segments.
For more information about actions on these nodes, see Customizing Each Segment.
Right-click the Interactive Modeling node to get these actions in the menu.
There is a limit of one Interactive Modeling node for each modeling node.
Right-click the Model Comparison node to get these actions in the menu.
After you choose a champion model, you can select the Select champion model option again and select Reset to restore the champion to the modeling node that was originally selected.
Click on the pipeline tab to perform these actions.
Runs the entire pipeline.
Stops a pipeline that is running.
Creates a duplicate pipeline. The name is appended with a number. You can rename the duplicate after it is created.
The pipeline for a segment of a pipeline cannot be duplicated.
Any changes you make to the settings in the Options pane are preserved in the duplicated pipeline.
Renames the pipeline.
Saves the pipeline with the nodes and any settings applied to those nodes as a template to The Exchange. The new templates can be used in other projects.
For projects that are imported from SAS Forecast Studio, the primary pipeline contains the Hierarchical Forecasting (Migrated) node. This pipeline should not be saved to The Exchange as a template. If this template is added to a project, the Hierarchical Forecasting (Migrated) node fails. That node is project-specific and works only in the project where it is first created.
Any changes you make to the settings in the Options pane are preserved in the template.
Deletes the pipeline from the project. At least one pipeline must exist in a project. If only one pipeline exists, this action is disabled.
Places a map of the pipeline in the upper left corner of the canvas.
Provides a space at the top of the pipeline to add a header. Use the header to provide a description or other text that might be useful. A toolbar is provided for rich text editing so that you can format the text in the header.

Copies the pipeline image for pasting into rich text applications, such as a word processor or email. The image does not paste directly into image editors. However, you can select it from the rich text application and then paste it into the image editor.
Copies the link to a pipeline in your project to your clipboard. You can share the link with other users who have access to the project. When the users open the link and sign in, they are directed to the selected pipeline in the Pipeline tab.
On the left side of the pipeline is the Nodes pane. This pane lists the modeling nodes that you can add to any pipeline.
If the Nodes pane
is not displayed, click on the left side of the pipeline to open it. You
can collapse the pane again by clicking
.
You can drag a modeling node over the Data node in the pipeline to add it. To remove a node, right-click the node and select Delete. Each pipeline requires at least one modeling node.
The Options pane on the right side of the pipeline displays options that you can set for a selected node in the pipeline.
If the Options pane
is not displayed, click to open it. You can collapse the pane again by clicking
.
After running a pipeline,
any subsequent changes to the options for a modeling node move the
pipeline back to Out-of-date state.
This is also true if you make a change and then change the setting
back to the original value.
When a pipeline is running, the options for each modeling node in the pipeline cannot be changed until the pipeline completes. If you need to make further updates to a modeling node after starting the pipeline, click Stop pipeline and wait for the pipeline to stop operation before you make any changes.