Binary numbers instead of one hot vectors

Question

While doing logistic regression, it is common practice to use one hot vectors as desired result. So, no of classes = no of nodes in output layer. We don\'t use

Answer 1

https://github.com/scikit-learn-contrib/categorical-encoding

Binary encoding (and in fact, base-anything encoding) is supported in category_encoders. In our case, we end up with feature per place in the binary string, so it's not one feature with value '011' or '010' its 3 with [0, 1, 1] and [0, 1, 0] respectively.