I'm not able to visualise what kind of 'trends' I would have to observe in multi-featured data to be able to say 'Logistic Regression would work well here'.

For example if I have only 1 feature and if the data is something like all negative data is class 0 and all positive data is class 1. Then I can clearly say that Logistic Regression works well here.

So what kind of 'analysis' of the data (multi-featured) would I have to do to decide whether Logistic Regression would work well or not?


1 Answer 1


As you would be aware that, logistic regression is a standard linear model, following these steps would be helpful :

  • Plot the input data, just observe how would probably the decision boundary look like.
  • If it's completely non-linear then other models will give you a good result than logistic regression.
  • If there is a high chance of linearity possible then you can go with logistic regression

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