Return quantity for demand sensing and forecasting

[nallurisai]

While performing the demand sensing forecasting using gradient boosting algorithm does it consider the return orders?

If yes is it a internal signal or external signal?

I am able to see only 4 internal signals in demand sensing with gradient boosting algorithm. what if we want to add few more internal signals to the algorithm. will the algorithm provide non-accurate results if we have the more than 4 internal signals?

As there is no option to add more internal signals how can we separate the return quantity from delivered quantity if we are using both the key figures as internal signals?

[LauraTozzo]

Hi Sai,

Demand Sensing considers both the "ordered quantity" and the "delivered quantity". If your return orders are somehow reflected in one of these two key figures, they are taken into consideration.

Regarding adding additional signals: you can add as many as you wish in the external signals section. The name of the section is only to distinguish between signals that are typically already defined in the system, but any key figure can be chosen as an external signal and it will be used by the algorithm when creating the forecast.

Best regards

Laura

[nallurisai]

Thank you Laura,

But I wanted to know that what if the returned quantities are not reflecting in either of the key figures.

How can a return quantity is added to demand sensing? Should it be created using custom key figures feature? If yes, how will it effect the calculation of the "demand sensing with gradient boosting" algorithm?

[LauraTozzo]

You can definitely add it by creating custom key figures. Demand Sensing with Gradient Boosting will then analyze the correlation between the newly added key figure and the main input key figure in the past and apply this knowledge to create the future forecast. Demand Sensing looks at each lag separately, so it could be that returned quantities influence some weeks more than others. If you want to understand better the influence of this key figure on the forecast you can use the importance or the impact key figures.