MODWT.transform

MODWT.transform(endog)

Decompose a series.

Runs the pyramid to the requested depth, inverts each level for the multiresolution components, and computes the unbiased wavelet variance with its degrees of freedom.

Parameters:

endog (ArrayLike) – The (T,) series, at least 32 observations.

Return type:

MODWTResult

Returns:

The MODWTResult.

Raises:

• DimensionError – If the series is not one-dimensional.

• SpecificationError – If the series is shorter than 32, or shorter than the requested depth allows.

Example

>>> rng = np.random.default_rng(0)
>>> y = rng.standard_normal(256)
>>> res = MODWT(levels=4).transform(y)
>>> res.coefficients.shape, res.scaling.shape, res.wavelet, res.nobs
((4, 256), (256,), 'la8', 256)
>>> MODWT(levels=6).transform(y)
Traceback (most recent call last):
    ...
cultivars.exceptions.SpecificationError: 6 levels of the la8 wavelet exceed what 256 ...