One-hot-encoder for categorical columns, whether predictor or response.
Multi-class prediction might presumably work with the existing LabelEncoder which makes it a single multi-level class. However, for predictors, this won't cut it unless the categorical variable is in fact ordinal.
When trying to find larger datasets for #92 there were none that only consisted of numerical predictors, hence we might need a one-hot-encoder if we are to use a larger dataset.
One-hot-encoder for categorical columns, whether predictor or response.
Multi-class prediction might presumably work with the existing
LabelEncoder
which makes it a single multi-level class. However, for predictors, this won't cut it unless the categorical variable is in fact ordinal.When trying to find larger datasets for #92 there were none that only consisted of numerical predictors, hence we might need a one-hot-encoder if we are to use a larger dataset.