In a Random Forest, I know that the Out Of Bag Error is described as the fraction of number incorrect classifications over number of out of bag samples.

Accuracy is defined as the number of correct classifications divided by the number of samples.

It seems to me like a good way to remember these is:

OOB Error = 1 - Accuracy

But, I just want to check before I teach this. Can someone please tell me what the difference is between these?

  • 1
    $\begingroup$ As standard for RF classifier, I would rather define OOB error as incorrect majority votes divided by the number of samples who at least once has been OOBsamples, typically equal to Ntraining samples, when number of trees is larger than a few. $\endgroup$ – Soren Havelund Welling Jan 18 '16 at 9:13
  • 1
    $\begingroup$ Yes you could define accuracy = 1 - OOB error $\endgroup$ – Soren Havelund Welling Jan 18 '16 at 9:15
  • 1
    $\begingroup$ I thought that too, but I think this page gives a good description of why that is not the case. stats.stackexchange.com/questions/198839/… $\endgroup$ – Alex Ilich Feb 9 '18 at 16:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.