1. Administering Planning
  2. Setting Up Predictions to Run Automatically with Auto Predict
  3. Creating Auto Predictions for Forecasts
  4. Configuring Advanced Auto Predict Options
  5. Data Screening Options

Data Screening Options

Historical data can have missing values and outliers, which are data points that differ significantly from the rest of the data, or can include events, which are typically one-off or recurring events that historically led to spikes or declines in the data. Data Screening options enable you to select several ways of handling missing values, identifying and adjusting outliers, and including events in predictions. Because adjusted outliers are treated as missing values, both of these situations are discussed and handled together.

Select from these options for Data Screening.

  • Include Events—This option is available for planners working with Predictive Planning on forms. When this option is selected, any events defined for the selected calendar are taken into consideration during the prediction. (Planners must first select a calendar. From the prediction area on the form, select Settings, and then select Date Ranges).

    If you include events, historical spikes or declines are also reflected in future predictions. For example, a North American calendar might include an event for Christmas, or an APAC calendar might include an event for Diwali, when sales would typically spike. By including the historical data spikes in the prediction, you see the spikes in the predicted data, so you can plan ahead for volume or to make use of the opportunity.

    Without events, spikes or falls in data are normalized and distributed over the prediction period, potentially leading to less accurate predictions.

  • Adjust Outliers—When this option is selected, when an outlier is detected in the series, outlier values are replaced with the prediction trend line value to avoid the impact of outliers.
  • Fill in missing values—When this option is selected, if there are missing values in the time series, the missing values are populated with the prediction trend line value to continue with the prediction.
  • Minimum missing threshold—When this option is selected, missing values in the time series are filled until the threshold is met. If the number of missing values is above the threshold provided, the prediction is not done. The maximum value can't be greater than 50%.