How You Forecast Product Returns

This topic explains how you can forecast product returns using forecasting profiles that are based on Bayesian machine learning.

If you must forecast product returns, Oracle recommends that you create two positive history streams, one for product demand and the other for product returns. Product sales and product returns don't necessarily occur in the same time buckets, so modeling two separate history streams is advisable. After you forecast the product demand and product returns with separate forecasting profiles, you can deduct the returns from the demand through a calculated measure.

You can import returns history only by using the file-based data import (FBDI) template for bookings history or shipments history. You can't collect returns history from other Oracle applications.

Note: Oracle doesn't recommend that you import negative values using the FBDI template for bookings history or shipments history to forecast product demand and product returns in one history stream. If you import negative values, they're set to zero during the forecasting process.