I have the problem, that a customer can buy something. Now I want to predict if the customer is buying another things in the next few days. So that you can summarize the packages and delivery not every package individually. Has someone read a similar issue? My data looks like

customerid order   article mandatoryDeliveryDate
1         03.05    Shoes   05.05
1         04.05    Paper   05.05
2         10.04    PS5     11.04
2         12.04    Laptop  16.04
3         28.04    Clock   30.04
3         [XXX]              

What I want to predict, if customer 3 is buying something in the next few days. Is it possible to predict if the customer 3 is buying something in the next few days taking into account the mandatory delivery date? And can you also give the probablity of these prediction?

Is there also any blog/paper/jupyter notebook or anything else where someone implemented a similar prediction for package delivery.


1 Answer 1


For clarity, let me state my understanding. The problem is about predicting whether a customer, after a 1st order, will put any additional order before a 'mandatory delivery date' when the 1st order must be mailed. The idea is if the customer does, we can group his multiple orders into a single package thus saving some mailing cost etc.

So the problem is: given some past data, predict if an event will happen in future. These kind of forecasting problems are very common, with many approaches exist.

Analysis e.g. association rules is one way; you may also phrase it into a typical classification problem by crafting some features on past data and label 0/1 target on whether an additional order happen. There are classical statistical approach e.g. time-to-event study too.

  • $\begingroup$ Thank you for your answer. Could I contact you about this? $\endgroup$
    – Test
    May 15, 2022 at 8:55

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.