2
$\begingroup$

I have a anomaly detection problem and my features are following exponential distrubition. Should I first transform my features into normal distrubition before feed into isolation forest?

$\endgroup$
1
$\begingroup$

Its better to use data as is for anomaly detection as the underlying data is not normal. Apart from what you asked, Isolation Forest has a problem that it has the chances of creating ghost clusters where anomaly scores would be incorrect. Hence the recent one is EIF works better.

$\endgroup$

Your Answer

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

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