mojolearn.RobustScaler#

class mojolearn.RobustScaler(*, with_centering=True, with_scaling=True, quantile_range=(25.0, 75.0), copy=True, unit_variance=False)[source]#

Bases: _PrepBase

Scale features by the median and the interquantile range.

Follows scikit-learn’s RobustScaler. Each column is centered by its median and divided by the range between the quantile_range percentiles, computed with numpy’s linear percentile over the non-NaN entries. NaN values are ignored during fit and kept as NaN by transform. A scale smaller than ten float32 epsilons is replaced by one. Input and output are float32.

Parameters:

  • with_centering (bool, default True) – Subtract the median before scaling.

  • with_scaling (bool, default True) – Divide by the interquantile range.

  • quantile_range (tuple of float, default (25.0, 75.0)) – (q_min, q_max) percentiles, with 0 <= q_min <= q_max <= 100; otherwise fit raises ValueError.

  • copy (bool, default True) – Accepted for scikit-learn compatibility; transform always returns a new array.

  • unit_variance (bool, default False) – Also divide the scale by norm.ppf(q_max) - norm.ppf(q_min), so that normally distributed data ends with unit variance. Requires 0 < q_min < q_max < 100 (ValueError otherwise).

Variables:

  • center_ (array of shape (n_features,) or None) – Per-column median, or None when with_centering=False.

  • scale_ (array of shape (n_features,) or None) – Per-column scale, or None when with_scaling=False.

  • n_features_in_ (int) – Number of columns seen in fit.

  • numeric_mode_ (str) – Numeric mode the estimator was fitted in; transform runs in the same mode.

Examples

>>> import numpy as np, mojolearn
>>> X = np.random.default_rng(0).random((1000, 8), dtype=np.float32)
>>> Xt = mojolearn.RobustScaler().fit_transform(X)
__init__(*, with_centering=True, with_scaling=True, quantile_range=(25.0, 75.0), copy=True, unit_variance=False)[source]#

Methods

fit(X[, y])

fit_transform(X[, y])

Fit to X, then transform X.

get_params([deep])

Return the constructor parameters as a dict.

set_params(**params)

Set constructor parameters.

transform(X)