Correcting outliers

When Planning Analytics Workspace identifies outliers in your data, you can adjust a forecast preview to correct the outliers.

About this task

When you choose to adjust for outliers in your forecast, the forecast model is corrected by using a normalized data point for the outlier to increase the forecast accuracy. The actual data point that is identified as an outlier is not modified on your Planning Analytics database.

Procedure

  1. Click the Outliers detected button on the preview.
    The Outliers detected pane opens.
    Forecast with outliers detected

    In this example, outliers are detected at the 2011 and 2012 data points.

  2. Click the Adjust outliers button on the Outliers detected pane.

    The outliers are adjusted, which has an impact on all following values. The Outliers adjusted pane explains the impact of the adjustment. In this example, the adjustment results in a 4% increase in the accuracy of the forecast.

    Forecast with outliers detected and adjusted

    When multiple outliers are detected, they are adjusted sequentially from the beginning of the data set. This means each adjusted outlier value is applied to the calculation of the adjustment for the subsequent outlier, resulting in a more accurate forecast.

  3. After you adjust outliers, an Outliers adjusted/Outliers not adjusted switch is available on the forecast preview. You can toggle the switch to see the forecast with or without the outlier adjustment. The Outliers adjusted/Outliers not adjusted switch is active on both the Preview chart and Statistical details tabs, so you can see the impact of outlier adjustment graphically and statistically.