Machine Learning – Hot encoding is not statistically ?friendly. I think you need the target to be a double normal distribution. Around 0 and around 1. For this you can add yt = yt +m ; m = model2(X) and then with subtract m from the model(X_test) – m.

Machine Learning – Hot encoding is not statistically ?friendly. I think you need the target to be a double normal distribution. Around 0 and around 1. For this you can add yt = yt +m ; m = model2(X) and then with subtract m from the model(X_test) – m.