I am a supply chain analyst working for a food industry. I have some knowledge in python, and forecasting but my job consists mostly in developing PBI dash and that is why I am far from being an expert in forecasting and I am requesting your help Today.
My problem looks quite simple. We have only one factory that is producing some food. We have some forecast made for us by a third party but its not very good and it is very expansive, that is why I want to develop an alternative in intern.
I want to predict future sales of food. I have all the sales orders since 2019 and I think I could easily use a SARIMA model since we have a very strong seasonality in our portfolio of products.
However, I also have already existing future sales orders at my disposal (e.g. we are in August, we already know most of the sales orders for September, and some for October.... December) and I would like to take them into consideration to make my forecast even more accurate. Of course, those future sales orders can change a lot. First, customers can cancel them, or move them to another month, but they also can ask for additional orders in a very short timeline, so this future data is not very accurate.
My question is how do I take those future sales orders in consideration when doing my forecast? Should I use a SARIMAX model and consider future sales orders as an exogenous variable? Should I use another model? I looked for a similar problem on Kaggle but could not find anything.
Thanks a lot for the guidance you could potentially offer.
