0
$\begingroup$

I understand that the general rule of thumb when working with dummy variables is to drop one column to avoid multicolinearity. The intercept term will take care of the reference dummy variable that's dropped.

However, what if I have multiple categorial variables. Say gender (Male vs Female) and occupation (employer vs employee). In this case, do I have two dummy variables (one for gender and one for occupation), or should I have four dummy variables (two for gender and two for occupation) and drop the intercept term instead?

$\endgroup$
2

1 Answer 1

0
$\begingroup$

The former. You are avoiding redundant information in your feature space construction. To adequately represent all your information you need on feature for gender and one for occupation.

$\endgroup$

Your Answer

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

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