Skip to content

On-premise users: click in-app to access the full platform documentation for your version of DataRobot.

Multiseries segmentation visual overview

Imagine that you sell avocados—different kinds (SKUs).

You want to predict avocado sales, so your target is Sales.

You sell these avocados in different stores, in different regions of the country. So your series ID is store.

Of course, stores sales don’t always have anything to do with one another. Maybe avocados sell often in hot places, and less often in cold places.

What you really need is a way to group series (stores in different regions) and forecast avocado sales based on that grouping. You can group the series ("stores") based on location and set that as the segment ID ("region").

Now you can build the right model for every segment, instead of one model for all. For example, you can model avocados that don’t sell very often with a Zero-Inflated XGBoost model.

You may even benefit from using a different metric per segment. Metrics are automatically selected based on target distribution.

How? Multiseries modeling with segmentation.

Updated December 12, 2021