What do we gain with one-hot encoding?

I don’t understand why is better to one-hot encode categorical data and use it as an input to logistic regression, for example. What is happening here?

If you use a numbered list, it can create a misleading linear relationship between the categories.

1 Like