When one-hot-encoding categorial features in python with pandas or sklearn, when should I drop one of the resulting columns? I recall something about having all columns present being a problem for machine learning algorithms because of multi-collinearity.

I did find 2 helpful questions here: Dropping one of the columns when using one-hot encoding Handling categorical variables in various ML algorithms

In my case, I'm one-hot-encoding for a random forest, so the answer to the 2nd post says I don't need to drop a column.

The answer to the first question seems to imply that you never have to drop a column because the software (i.e. python in my case) will handle multi-collinearity issues.

Can someone confirm if that is true and shed light on when you would drop one-hot-encoded columns in python?


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.