I understand there are lot of ML products that can help predict the time to event. For ex, customer will purchase in the next 30 days or Patient has chances of re-admission in next 60 days etc.

But why do you think why don't people predict to the level of date?

ex: For ex, customer will make his next purchase on Feb 2nd, 2016 or patient has chance of readmitting on Mar 23rd, 2018.

Don't know from an ordinary layman perspective, I feel dates might be more useful. Of course, all our predictions are reported based on confidence Intervals.

For ex: If we know the date when the course will end, we can plan our schedule accordingly to do other tasks/sign up for new courses etc. But if we are told that your course might end anytime in next 30/60/90 days, we are at risk of wasting few days before we enroll for another course etc. Hope my example makes sense.

Can experts here suggest me on when

a) Prediction to date level might be a good thing?

b) Prediction to date level might be a bad thing?


1 Answer 1


The reason you get predictions without an exact time is because that is how models are trained. They are not trained to predict the exact time but a window for it.

The reason a model is not trained to predict exact time is because it introduces a lot of problems starting with data imbalance and the huge number of classes it would introduce. Also, DL/AI is not God or Laplace's demon, so to predict the exact is not going to happen.


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.