CODEWITHSUNDEEP
machine-learningclassificationlogistic-regressionfeature-extractionfeature-detection
BlueGirl
BlueGirl

Reputation: 501

Mixing binary and non-binary features in logistic regression

I have some binary features (0 or 1) and some non-binary features which have values between 0 and 1 (such as 0.24). I use Weka logistic regression to classify instances with all these features. Does it work properly?

Thanks

Upvotes: 0

Views: 832

Answers (1)

lejlot
lejlot

Reputation: 66805

The question of how to efficiently combine categorical and numeric features is an open research question. However, in your case, where you already decided to encode categorical as 0/1 and remaining as x e [0, 1] there is nothing to "worry" about. This is a valid application of logistic regression. Simply bare in mind that there are no guarantees that this is the best way to represent your data. It will work, however some specific weighting can be better, but this is purely data dependent property, which cannot be easily answered.

Upvotes: 1

Related Questions