Running the Forecast Pipeline

Before running a pipeline, be certain that your settings for variable assignments and project settings are correct. After you run one or more pipelines successfully and open the Overrides tab, many of these settings are committed and cannot be changed. Before running the complete pipeline, you can run just the data node and review the results.

1. Right-click the Data node and select Run.

   When the Data node runs, it performs these tasks with the project data:

   • accumulates the input time seriesan aggregation of transactional data into specified time intervals and sorted according to unique combinations of the default attributes (BY variables) based on the Time interval value set for the time variable
   • interprets missing values based on the Missing interpretation value set for the dependent variable
   • generates the attributes for descriptive statistics
   • generates the attributes used for filtering and segmenting project data based on demand classification
   • merges the attributes from the descriptive statistics and demand classification with the attributes from the input data for the project

   If the input data for the project is very large, this process can take a long time.

2. When the Data node has finished processing successfully, right-click the Data node and select Time series viewer.

The Time Series Viewer shows the plot of the historical data. The data is accumulated using the time interval settings that you specified in the variable assignments. The project attributes are listed on the left. If you need to make any final changes to these settings, make those changes before running the full pipeline.

When you are certain that the variable assignments and project settings are good, follow these steps.

1. Select the Model Comparison node and examine the selection criteria in the options panel on the right. These statistics of fit are used to choose the best performing modeling node in the pipeline. For more information, see Weighted Model Comparison Selection Criteria.
2. Click Run pipeline. The pipeline starts with the Data node and is finished after the Output node is complete.

   While the pipeline is running, you can create another pipeline or exit the project. If you remain on the Pipelines tab long enough, a message is displayed when the pipeline completes.

3. Right-click the Model Comparison node and select Results. The results page shows a table with all of the nodes in the pipeline. The node with the best results is designated as the champion for this pipeline.

If there are any problems running the pipeline, check the logs of the modeling nodes. To open the log, right-click a modeling node and select Log. The log shows the code for the node along with any notes, errors, and warnings.

See Also

• Creating Your First Project
• Model Comparison
• Pipeline Fails to Run for Project with User-Defined Formats