1
$\begingroup$

I have 100s of columns with binary values [0, 1] plus some extra columns without binary values. I am trying to do regression model but the model performance is very low. For non-binary features, I have used PCA to decrease the dimension of it. I don't think its appropriate to use PCA for binary values. I am guessing, its because of the large number of binary columns, the model isnt doing great. What can be done on this kind of situation ? I have tested with almost all of the regression model available in sklearn.

What approach I could take to improve the model performance ? Any suggestions.

$\endgroup$

1 Answer 1

1
$\begingroup$

An approach worth exploring for dimensional reduction of these binary features would be Multiple Correspondence Analysis (Python libraries such as Prince and mca are available), which is similar to PCA but deals with nominal categorical variables. Here is an excellent answer on applicability of PCA and factor analysis to binary features and another one on the same topic.

$\endgroup$

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.