The following pipeline templates are provided by SAS Visual Forecasting. Some of these templates have specific project requirements that cannot be established when a new project is created. For this reason, make sure you read these descriptions before selecting a pipeline template for a new project.
These templates are good to use for any forecasting project using a time seriesan aggregation of transactional data into specified time intervals and sorted according to unique combinations of the default attributes (BY variables) data source. You to compare the different templates and other modeling nodes to find the template that works best for your data.
This template simply creates a pipeline with a single Data node. You need to add at least one modeling node to complete the pipeline. If you create a project selecting an invalid pipeline template, the project is created using the Base Forecasting template.
This template is valid only for external forecasta numerical prediction of a future value for a specified time period for each unique combination of BY variable values projects. This template is automatically loaded as the pipeline for any external forecast projects that you create. The schema for external forecasts is explained in Working with External Forecast Projects.
If a new time series project is created with this template, the project uses the Base Forecasting template instead.
These templates are intended for forecasting projects that include established segmentation definitions.
This template is valid only for projects with defined segments. If you specify this template during project creation, SAS Visual Forecasting cannot determine the segmentation definitions and the default pipeline is set to use the Base Forecasting template instead.
For more information, see Defining Project Segments.
This template segments the time series based on demand classification patterns. The classification is not performed until after project creation is complete. See Demand Classification for more information about this pipeline template.
These templates are intended to be used for data that has been segmented based on demand patterns in the time series. When using a Demand Classification pipeline, these templates are selected for use in nested pipelines within each segment, based on the classification for that segment. Some of these templates might be suitable for time series projects. Other templates are not. For example, Retired Forecasting is intended for time series with large periods of inactivity and does not produce any forecasts or measurements for statistics of fit.
The following templates are assigned to each demand class in a project.
|
Demand Class |
Pipeline Tempate Assignment |
|---|---|
|
INSEASON_INTERMITTENT |
Regression Forecasting |
|
INSEASON_NON_INTERMITTENT |
Regression Forecasting |
|
LOW_VOLUME |
Naive Forecasting |
|
OTHER |
Naive (Moving Average Forecasting) |
|
RETIRED |
Retired Forecasting |
|
SHORT |
Naive (Moving Average Forecasting) |
|
YEAR_ROUND_INTERMITTENT |
Auto-forecasting (Intermittent) |
|
YEAR_ROUND_NON_SEASONAL |
Non-Seasonal Forecasting |
|
YEAR_ROUND_OTHER |
Naive (Moving Average Forecasting) |
|
YEAR_ROUND_SEASONAL |
Seasonal Forecasting |
|
YEAR_ROUND_SEASONAL_INTERMITTENT |
Temporal Aggregation Forecasting |
For a description of each demand class, see Detailed Description of Demand Classifications.