ValueClipper transformer.
ValueClipper values to a specified range.
Classes
Clip feature values to a specified range or percentile.
Constrains the numerical range of features by capping values at the specified lower and upper thresholds.
Constructor
__init__( self, lower: Optional[float] = None, upper: Optional[float] = None, percentile: Optional[tuple] = None, columns: Optional[List[int]] = None, )
Parameters
lower
float
Fixed lower bound. Values below this are set to
lower.
upper
float
Fixed upper bound. Values above this are set to
upper.
percentile
tuple of float
Percentile boundaries expressed as
(lower_pct, upper_pct). If provided, the absolute bounds are calculated from the training data (e.g., (1, 99) for winsorization).
columns
list of int
Indices of columns to clip. If
None, all columns are processed.
Attributes
bounds_
dict
Calculated (lower, upper) boundaries used for each column.
Clip values to the [0, 10] range:
python
>>> from tuiml.preprocessing.outliers import ValueClipper
>>> import numpy as np
>>> X = np.array([[-10], [5], [20]])
>>> clipper = ValueClipper(lower=0, upper=10)
>>> X_clipped = clipper.fit_transform(X)
>>> print(X_clipped.flatten())
[ 0. 5. 10.]Methods
get_parameter_schema
(cls)
fit
(self, X: np.ndarray, y: Optional[np.ndarray]=None, feature_names: Optional[List[str]]=None) -> 'ValueClipper'
transform
(self, X: np.ndarray) -> np.ndarray
__repr__
(self) -> str