Many significant enhancements have been made to SAS Visual Forecasting over the past year. The improvements highlighted in this article are those introduced in 2025 and relate to the following nodes: Auto-forecasting (AF), Hierarchical Modeling (HM), Save Data, and Interactive Modeling (IM).
Users can now import external models into the AF node. These imported models are treated as system-generated models, applied to all series in the node, and included in the champion tournament selection process.
From the HM node properties panel, users can modify the reconciliation level and method, which is inherited from the default values specified on the Data tab. Recall that users can also set the default pipeline to be used for each level of the hierarchy in the properties panel.
When it comes to pipeline management, users can select a different pipeline from The Exchange for one or more levels of the hierarchy within the HM node. And users can reset the pipeline to the default for one or more hierarchy levels.
Users can also view reconciliation information in the facet panel and use that information within filters on the Forecast viewer and in Overrides.
From the Save Data node following an HM node, users can view and save output tables at each level of the hierarchy. The output tables are the
- Pre-reconciled forecasts
- Reconciled forecasts
- Pre-reconciled statistics
- Reconciled statistics
When creating system-generated models within the IM node, users can specify which model families they want included (ESM, ARIMA, etc.) in the model generation process.
Within the IM node, users can create a zero or missing forecast for a time series in the Interactive Modeling node. This is useful when a product is discontinued (either permanently or temporarily), but you want to keep the history for hierarchical modeling.
In addition, users can share models across time series from within an IM node. Users can add models to the project model repository from one series and then copy models from the repository to another series. All user-saved models to/from the repository are treated as custom models and are tear-off copies. This means that there is no linkage between the models in the repository and the IM node – they are independent of one another. And there are 59 predefined models provided by SAS.
After creating custom models, users can select and delete one or more custom models that do not perform well. A copy of the model is created when a system-generated model is made the champion or when the user manually creates a copy from a system-generated model. In both cases, these copies are considered custom models.
When switching between BY groups in the IM node, the champion model remains the default selection and is shown at the top of the model list.
A Constituent Node Weight (or OUTWEIGHT) table can be generated in the Interactive Modeling node. Users can view the contents in the Model Comparison node if the Interactive Modeling node is the champion modeling node. Otherwise, users can access the table from a Save Data node placed after the Interactive Modeling node.
The IM node also contains an option to enable a hierarchy, specify a reconciliation level and method, and provides a tree structure to navigate the hierarchy. The hierarchy in the IM node can be enabled even when the parent node(s) are not hierarchy-based modeling nodes.
Other features that have been added but will not be covered in the demo include:
- Users can reopen the "Series Update Status" dialog from anywhere within the VF project.
- Users can "Modify Event Usage" via a right-click option to the Temporal Aggregation (TA) node.
- The Series Analysis plots/tables layout is now persisted, by user, in the Series Analysis view within the Interactive Modeling node. The layout persists when you move from one series to another and after closing either the node or the project.
In the demo below, I’ll share with you how to use these features. Happy Forecasting!