How to formulate a random event and build a Machine Learning algorithm?

I am working on a problem and wanted to ask your ideas/suggestions. Let' suppose you have a vendor who over- supplies materials to you on and off. What I mean is that sometimes they provide enough materials sometimes they over-provide. It does not happen on a continuous basis and it is sporadic. And of course, you are being charged for this over-supply. You have enough data to work with over a 2 years time frame. Now I want to build ML model that is able to forecast this over-supply when a new order comes onsite. I have a dataset ready for this. You know that this is not a continuous variable where you could use linear/non-linear relationships to predict. Since I know the tolerance for each order I can assign labels Yes/No to my data. Do you think a classification algorithm would be enough?