OneHotEncoder transformer.
One-hot encoding for categorical features.
Classes
One-hot encoding for categorical (nominal) features.
Transforms categorical features into a binary vector representation where each category is represented by a separate column.
Constructor
__init__( self, categories: Optional[List[List]] = None, drop: Optional[str] = None, columns: Optional[List[int]] = None, )
Overview
One-hot encoding is essential for algorithms that cannot handle categorical data directly. It creates N binary columns for a feature with N unique categories (or N-1 if dropping a category).
Parameters
categories
list of list
Manually specified categories for each column. If
None, categories are inferred from the training data.
drop
{"first", "if_binary"}
Strategy to drop one category per feature to avoid multicollinearity:
- •
None: Keep all categories. - •
"first": Drop the first category. - •
"if_binary": Drop the first category only if the feature is binary.
columns
list of int
Indices of columns to encode. If
None, all columns are encoded.
Attributes
categories_
list of np.ndarray
The categories determined during fitting for each encoded column.
Encode a single categorical feature:
python
>>> from tuiml.preprocessing.encoding import OneHotEncoder
>>> import numpy as np
>>> X = np.array([[0], [1], [2], [0]])
>>> encoder = OneHotEncoder()
>>> X_encoded = encoder.fit_transform(X)
>>> print(X_encoded.shape)
(4, 3)Methods
get_parameter_schema
(cls)
fit
(self, X: np.ndarray, y: Optional[np.ndarray]=None, feature_names: Optional[List[str]]=None) -> 'OneHotEncoder'
transform
(self, X: np.ndarray) -> np.ndarray
get_feature_names_out
(self, input_features: Optional[List[str]]=None) -> List[str]
__repr__
(self) -> str