Hi everyone,
I noticed that when generating a baseline forecast using Azure machine learning, it tends to ignore the zeroes in the historical demand. in the figure below, the product has a historical demand for 6 pcs for 12/1/2020, but when the forecast is generated, the system came up with 6 pcs forecast every month.
Figure 1. Historical demand for 6 pcs only for 12/1/2020
Figure 2.Generated forecast using Azure ML based on the historical demand showing 6 pcs forecast every month
The resulting calculation is far from reality and is not very accurate. Is there a way for AML to consider the months where historical demand is zero? Also, the client doesnt want to use Copy over historical forecast as alternative forecast generation strategy.
Thank you everyone and stay safe.
Best regards,
Eric
