Running Auto Predict Predictions

You can run an Auto Predict prediction definition from the Auto Predict page or the IPM page, or you can schedule a job to run the prediction automatically.

Running an Auto Predict Prediction

To run an Auto Prediction prediction definition:

  1. From the Home page, click Application, then click Overview, and then from the Actions menu, select Auto Predict.

    Alternatively, if you are using IPM Insights, from the Home page, click IPM and then click Configure.

  2. From the list, click Actions next to an Auto Predict definition, and then select Run.

    For Auto Predict predictions with a large number of cells to calculate, in Run Parameters, select a dimension and then click Run Now.

    This setting is used to speed up an Auto Predict job by running predictions in parallel in separate prediction threads. For these parallel jobs to be efficient, choose a dimension that will result in evenly spread data for each prediction thread.

  3. To review a summary of the prediction, review details in the Jobs Console: Click Application, then Jobs, and then click the prediction definition name on the Jobs page.

If a long-running prediction job is interrupted by an automated maintenance window, when the maintenance is complete, the prediction job picks up where it left off.

Scheduling an Auto Predict Job

For information about scheduling Auto Predict jobs, see Scheduling Jobs. Auto Predict has its own job type, called Auto Predict.

For Auto Predict predictions with a large number of cells to calculate, in Run Parameters, select a dimension and then click OK. This setting is used to speed up an Auto Predict job by running predictions in parallel in separate prediction threads. For these parallel jobs to be efficient, choose a dimension that will result in evenly spread data for each prediction thread. For recurring jobs, the selected dimension is used for jobs until you change it.

Before running a scheduled prediction job, Auto Predict analyzes certain factors to determine if prediction results would change. If the prediction results would not change, the Auto Predict job is not run. For example, if the historical range was set for the last three years and there has been no change in the start period/year for prediction, the historical data for analysis would be the same and the prediction results wouldn’t change. In this case, to optimize performance and prevent unnecessary jobs from running, the Auto Predict job is skipped. A message is displayed indicating that the job will not run. If you want to run the prediction anyway, you can run the job from the Auto Predict page.