Skip to content

Product metric #689

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
luisheb wants to merge 18 commits into
base: develop
Choose a base branch
from
Open

Product metric #689

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
18 commits
Select commit Hold shift + click to select a range
4407c7c
pproduct metric class
Apr 22, 2025
46a7e93
runtime_checkable in metric
Apr 22, 2025
c2d1cb1
pairwise optimization
Apr 23, 2025
17ba36e
small correction
Apr 24, 2025
9281a5a
Test pproduct metric first version
Apr 24, 2025
95309cf
DataFrame corrections
Apr 25, 2025
c83cc47
Pairwise metric done + Correction of errors. Ready
Apr 26, 2025
4811d15
mypy fix
May 5, 2025
276f37f
ruff corrections
May 6, 2025
bdd536e
test and doc pproduct
May 8, 2025
f7c0e92
shape correction
May 9, 2025
9320237
product metric compatible with grid search
May 13, 2025
e5500b9
Merge branch 'GAA-UAM:develop' into feature/product_metric
luisheb Jun 16, 2025
45255a2
Merge branch 'develop' into feature/product_metric
luisheb Jun 18, 2025
adbd23e
Ruff and tests
Jun 18, 2025
909ca4b
doc
Jun 18, 2025
b3fe8ac
Tests raise error
Jun 18, 2025
aed2500
type for dispatch
Jun 18, 2025
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
36 changes: 36 additions & 0 deletions docs/modules/misc/metrics.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -110,3 +110,39 @@ transformation followed by another metric:
:toctree: autosummary

skfda.misc.metrics.TransformationMetric


Product metric for mixed data
-----------------------------

In applications involving heterogeneous data—such as combinations of
scalar values and functional observations—it is often useful to define
a metric that captures distances across all components in a consistent
way. The following class implements a weighted product metric that
supports combining multiple data types, each potentially requiring
a different metric.

This class generalizes the idea of computing a norm over a product
space, where each component may be associated with a different scale
or importance. It supports common functional data representations
(:class:`~skfda.representation.FDataGrid`,
:class:`~skfda.representation.FDataBasis`), numeric arrays, and
:class:`pandas.DataFrame` objects mixing them.

.. autosummary::
:toctree: autosummary

skfda.misc.metrics.PProductMetric
skfda.misc.metrics.pproduct_metric
skfda.misc.metrics.DefaultMetric

The metric is defined as a weighted :math:`L^p`` norm of component-wise
distances. Each component may be assigned its own metric and weight,
allowing fine-grained control over the overall distance calculation.
This is especially useful in machine learning tasks that involve mixed
input types, such as classification or clustering over functional and
scalar features.

Functional wrappers for ease of use are also provided, along with a
default metric that infers the appropriate behavior depending on the
input type.
12 changes: 12 additions & 0 deletions skfda/misc/metrics/__init__.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,6 +36,12 @@
"TransformationMetric",
"pairwise_metric_optimization",
],
"_pproduct_metric":[
"PProductMetric",
"pproduct_metric",
"DefaultMetric",
"default_metric",
],
},
)

Expand Down Expand Up @@ -63,6 +69,12 @@
)
from ._mahalanobis import MahalanobisDistance as MahalanobisDistance
from ._parse import PRECOMPUTED as PRECOMPUTED
from ._pproduct_metric import (
DefaultMetric as DefaultMetric,
PProductMetric as PProductMetric,
default_metric as default_metric,
pproduct_metric as pproduct_metric,
)
from ._utils import (
NormInducedMetric as NormInducedMetric,
PairwiseMetric as PairwiseMetric,
Expand Down
Loading