multi_resolution

Defined in fynance.features.engineering

multi_resolution(func, X, windows, **kwargs)

Stack a window-based feature computed at several resolutions.

Applies func(X, w, **kwargs) for every window w in windows and column-stacks the results, letting a model learn the relevant horizon instead of fixing one.

Parameters:

funccallable

A feature function taking (X, w, **kwargs) (e.g. sma, realized_volatility).

Xnp.ndarray

One-dimensional input series.

windowsiterable of int

Window sizes / resolutions.

**kwargs

Extra keyword arguments forwarded to func.

Returns:

np.ndarray

Array of shape (len(X), len(windows)), one column per resolution.

Examples

>>> import numpy as np
>>> from fynance.features.momentums import sma
>>> X = np.arange(1., 6.)
>>> multi_resolution(sma, X, [2, 3]).shape
(5, 2)