I m making my first steps in AI and ML. I choose myself a project, I want to fix with ML, but I m unsure which methode to use.
Business Case: A Customer can put offers and set a date he wants to receive his products. He is able to change the amount of products he buys at every time. I have to deal with the costs of unbuyed products and missing profit, in case I produced less than he wanted. I have plenty of data from past transactions contianing the original amount of products ordered and the amount I sent to the costumer. My goal is to get a predicitve analytics model which is able to tell me after a costumer changed the number of products from an order, how probably this change is final.
I m really new to this topic and are not quite getting all the information for the different methodes. I know classification and regression are the big players and can be implemented in different ways. But is one of those approaches fitting for my problem?
Many Thanks in advance.
You can go with a classification based approach. Since you goal is to predict whether the order change is final or not. The probability of happening that change can be derived from the accuracy/F1 score of your model. Higher the values, higher successful predictions. In laymen's terms think this as classifying whether the order is final or not.
You have to go for a regression approach if you're trying to predict a value based on the order change. For example if you want to predict what is the cost for the next order change, then you have to use regression.
As I understood your use case matches with the first scenario.