Dataset : www.kaggle.com/c/cat-in-the-dat-ii/data
It seems that NN are the real way to go.
Target Encoding outperforms One Hot Encoding and Label Encoding.
sumansahoo16/Categorical-Feature-Encoding-Challenge-II
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.