lazymatrix.standardize

LazyMatrix for Python · 0.1.1

Source

standardize(
    X: typing.Any,
    *,
    center: bool = True,
    scale: bool = True
) -> lazymatrix.LazyMatrix

Fit column means and population scales and return a lazy matrix.

Parameters

X : numpy.ndarray or scipy.sparse matrix or array

Two-dimensional real numeric input. Sparse input must use CSR or CSC. The returned matrix owns a copy of the raw values in float64 storage.

center : bool, default=True

Subtract each column's mean during operations.

scale : bool, default=True

Divide each column by its population standard deviation, replacing exact zero scales with one.

Returns

LazyMatrix

A matrix retaining fitted column parameters without constructing a normalized copy. Statistics are fitted immediately.

Raises

TypeError

If the input type, storage format, values, or options are unsupported.

ValueError

If the input is not two-dimensional or has no rows.

Notes

This first release fits each input independently. Reusing fitted normalization on new observations is not yet exposed by the Python API.

Examples

>>> import numpy as np
>>> from scipy import sparse
>>> from lazymatrix import standardize
>>> Z = standardize(sparse.csr_matrix([[1.0, 0.0], [3.0, 4.0]]))
>>> Z @ np.ones(2)
array([-2.,  2.])