I have data for of past flights with quite a lot of features. I am trying to think whether a logistic regression would be suitable to predict which type of flights are likely to be late in the future. 15 or more minutes.

Since I only need to find out whether or not they are likely to be late (binary outcome, yes or no), then I figured that a logistic regression model could be suited to this problem.

However the past data is not binary, I have lateness (in minutes). Is that alright? Can I still build the model?

P.S. I haven't yet built the model, I'm currently in the planning stage and was wondering whether or not it's worth doing!

  • 1
    $\begingroup$ LR can work here. You can transform lateness in minute into a new target : "is it later than 15 minutes?" Yes or No. You also have to do EDA, check and maybee do features reduction. Then trying different models can be interesting. $\endgroup$
    – Malo
    Commented Jun 8, 2021 at 22:48
  • $\begingroup$ @Malo Yeah that's a good idea! Thanks! $\endgroup$ Commented Jun 8, 2021 at 23:02


Your Answer

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

Browse other questions tagged or ask your own question.