Change Points Considered by Demand Sensing with Gradient Boosting

You can now set the demand sensing with gradient boosting algorithm in the Manage Forecast Models app to consider change points that were identified in the time series by the last forecast automation job. When you do so, the system generates a signal for these change points. The values of this signal are added to the input data along with the calendar signals and open orders to help the algorithm calculate a more accurate forecast.

In general, there are two types of change points that may occur in a time series - trend changes and level shifts. Demand sensing with gradient boosting can currently only consider level shifts.

Once you enable the new option for a forecast model, you can also specify an importance key figure for it.