Is it possible to build a propensity model (i.e., the likelihood that a user will buy an item) using only positive values.

For example, I have a bunch of data about Customers (people that bought stuff) and Users (people that haven't bought stuff yet) I want to get the likelihood that a User becomes a Customer.

It seems that the only way to do so is to train a model using the data of Customers, therefore using only Positive values.


2 Answers 2


The model built using only positive value will be biased towards positive value and will always predict for the customers who bought any items before. But customers who didn't buy anything, the model won't be able to predict anything.

  • $\begingroup$ What about one class classification models? I have read that Positive and Unlabelled learning is a hot topic right now. Thoughts? $\endgroup$
    – QuantNoob
    Jun 21, 2020 at 15:22
  • $\begingroup$ If you have only one class then what is the meaning of classification problem. Unlabeled training is done in unsupervised learning where you have only input data but no output and all you do is gain insight from the input. K Means Clustering is one of them. You can check that. $\endgroup$
    – SrJ
    Jun 21, 2020 at 15:25

The type of model depends on the data.

If the data includes the time point that some people converted, then analysis can identify the features associated with conversion. One option is Kaplan–Meier estimation for conversion / not conversion.

If the data does not have the time of conversion but still each row is labeled as customers / not customers, then it can be framed as a binary classification. Any traditional classification model could be used, popular models include boosted trees and linear models.

If the data is only customers, then one-class classification can be performed.


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.