Suppose that I possess a binary classification model, where the positive class is certain, and I am confident that the positive samples belong to that class, while the negative class is uncertain, as I do not have any prior knowledge of whether these samples belong to the positive class or not. Is there a specific term used to refer to such a classification model?

Example: We have a group of individuals who are gamers. Some of my clients are gamers with certainty, as I can see their transactions and purchases on Steam, Nintendo, or Microsoft Games. For the rest of my population, I am unsure whether they are gamers or not, as it is possible that they are gamers but are not visible in my transactional data.

The objective of the model is to calculate the probability of a person being a gamer for those clients for whom we do not have certainty.

Does this kind of model in which there is certainty for a group and uncertainty for the other group have a special name or is it just plain binary classification?


1 Answer 1


This seems very close to one-class classification, where one only has training data for the positive class (the "certain" class) but the goal is for the model to distinguish positive vs. negative. In other words, the model can only create a model of the positive class and then classify instances based on whether they match the pattern of this class.

  • $\begingroup$ could it be semi-supervised learning too? $\endgroup$ Commented Feb 28, 2023 at 0:23
  • $\begingroup$ @JuanEstebandelaCalle Yes it's certainly relevant, but semi-supervised learning does not involve that one specific class is known vs. the other unknown. I would assume that semi-supervised methods consider that in the few labelled instances there are cases of the two classes. If correct, this could be an issue for your case. $\endgroup$
    – Erwan
    Commented Feb 28, 2023 at 16:35

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.