TransformationMetric#

class skfda.misc.metrics.TransformationMetric(transformation, metric)[source]#

Compute a distance after transforming the data.

This is a convenience function to compute a metric after a transformation is applied to the data. It can be used, for example, to compute Sobolev-like metrics.

Parameters:

e1 – First object.
e2 – Second object.
transformation (Callable[[Original], Transformed]) –
metric (Metric[Transformed]) –

Returns:

Distance.

Examples

Compute the L2 distance between the function derivatives.

>>> import skfda
>>> from skfda.misc.metrics import l2_distance, TransformationMetric

>>> x = np.linspace(0, 1, 1001)
>>> fd = skfda.FDataGrid([x], x)
>>> fd2 = skfda.FDataGrid([x/2], x)

>>> dist = TransformationMetric(
...     transformation=lambda x: x.derivative(),
...     metric=l2_distance,
... )
>>> dist(fd, fd2)
array([ 0.5])

Methods