With Auto Predict, administrators can define a prediction to predict future performance based on historical data and schedule a job to run that prediction definition, automating the prediction process.
Auto Predict uses the same forecasting and statistical methods as Predictive Planning, but unlike Predictive Planning, you aren’t limited to running a prediction on a form. You can predict values for thousands of cells at once, when needed, by scheduling a job, and automatically write prediction results to a scenario/version—in the same cube or in a different cube from the historical data. You can also include Best and Worst Case prediction results. When you run the prediction, historical data for each member in the Auto Predict definition is retrieved and then analyzed using time series forecasting techniques to predict the future performance for these members.
Auto Predict is useful:
When you have large amounts of data to predict.
To jumpstart the forecasting process by pre-filling a prediction scenario with predictions based on historical data. Then, planners can compare predictions to forecasts.
For variance analysis. Seed a prediction scenario with prediction results, and compare predictions to forecast. To incorporate variance thresholds, you can define a groovy rule that requires planners to add explanatory commentary at a defined threshold, for example if the forecast is 5% greater than the prediction.
To pre-populate your Forecast or Plan scenarios before starting a forecast or planning cycle. Planners can use these prediction results as a starting point for their forecasts, and make adjustments as needed.
To keep predictions up-to-date as actuals come in by scheduling Auto Predict jobs. When new actuals come in, the job can run automatically to populate the cube; planners can see updated predictions and compare to forecast and plan.
Use Auto Predict results in forms, dashboards, and in ad-hoc analysis. You can also use Auto Predict prediction results within Smart View.
You define a prediction to read historical data from a cube (typically an ASO cube) and write prediction results to the same cube or to a different cube, for example to a BSO cube where planning and forecasting is done.
Auto Predict works with EPM Standard and EPM Enterprise applications for Custom and Module application types when Hybrid Essbase is enabled. For legacy applications, Auto Predict works with the Enterprise application type when the Essbase version is upgraded to the version that supports Hybrid Essbase and Hybrid Essbase is enabled.
Auto Predict works with custom time periods in Planning and alternate time periods in Planning modules - for example in Financials you can predict at both a weekly level and at a monthly level in the same application.
Prediction results are more accurate the more historical data you have. There should be at least twice the amount of historical data as the number of prediction periods.
When series have less than six historical data points, Auto Predict does not use forecasting, instead it uses a straight line fit for the prediction series. In this case there are results only for the base case. Best case and worst case are not calculated, because the values would be the same as the base case.
- For details about the statistical forecasting methods used in Auto Predict predictions, see Predictive Planning Forecasting and Statistical Descriptions in Working with Planning.
- Planners can do predictions directly on a form. For more information, see Improving Forecasting Accuracy with Predictive Planning in Working with Planning.
To work with Auto Predict:
- Define the Auto Predict prediction. See Creating Auto Predictions for Forecasts.
- Modify application settings if needed to set the Prediction Interval. These settings are shared with Predictive Planning. By default, the prediction interval for worst and best cases is 2.5% and 97.5%. See Managing Application and System Settings.
- Run the Auto Predict prediction: See Running Auto Predict Predictions.
- Run the Auto Prediction from the Auto Predict page.
- Schedule a job to run the Auto Predict prediction automatically.
|Your Goal||Watch This Video|
Learn how to set up an auto predict definition, and schedule a job to run that definition. See how to prefill a prediction scenario to jump start the forecasting process, and how to directly compare the system’s predictions to your ongoing forecast in real time. You compare predictions to Forecast and incorporate variance thresholds that require you to enter explanatory commentary. Also see how to seed plans or forecasts prior to starting the forecast cycle.