Skip to content

add noise_variance computation from cpp side #2371

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

Draft
wants to merge 2 commits into
base: main
Choose a base branch
from
Draft

add noise_variance computation from cpp side #2371

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
2a912ba
minor fix
Alexandr-Solovev Mar 21, 2025
0f6f8c2
minor fix
Alexandr-Solovev Mar 21, 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
4 changes: 4 additions & 0 deletions onedal/decomposition/pca.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -112,6 +112,10 @@ void init_train_result(py::module_& m) {
.def_property_readonly("explained_variances_ratio",
&result_t::get_explained_variances_ratio)
#endif // defined(ONEDAL_VERSION) && ONEDAL_VERSION>=20240100
#if defined(ONEDAL_VERSION) && ONEDAL_VERSION >= 20250100
.def_property_readonly("noise_variance",
&result_t::get_noise_variance)
#endif // defined(ONEDAL_VERSION) && ONEDAL_VERSION>=20250100
.def_property_readonly("means", &result_t::get_means)
.def_property_readonly("variances", &result_t::get_variances);
}
Expand Down
18 changes: 1 addition & 17 deletions onedal/decomposition/pca.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -104,22 +104,6 @@ def _resolve_n_components_for_result(self, shape_tuple):
else:
return self.n_components

def _compute_noise_variance(self, n_components, n_sf_min):
if n_components < n_sf_min:
if len(self.explained_variance_) == n_sf_min:
return self.explained_variance_[n_components:].mean()
elif len(self.explained_variance_) < n_sf_min:
# TODO Rename variances_ to var_ to align with sklearn/sklearnex IncrementalPCA
if hasattr(self, "variances_"):
resid_var = self.variances_.sum()
elif hasattr(self, "var_"):
resid_var = self.var_.sum()

resid_var -= self.explained_variance_.sum()
return resid_var / (n_sf_min - n_components)
else:
return 0.0

def _create_model(self):
# Not supported with spmd policy so BasePCA must be specified
m = BasePCA._get_backend(BasePCA, "decomposition", "dim_reduction", "model")
Expand Down Expand Up @@ -185,6 +169,7 @@ def fit(self, X, y=None, queue=None):
self.explained_variance_ratio_ = from_table(
result.explained_variances_ratio
).ravel()
self.noise_variance_ = from_table(result.noise_variance_).ravel()
self.n_samples_ = n_samples
self.n_features_ = n_features

Expand All @@ -194,7 +179,6 @@ def fit(self, X, y=None, queue=None):

n_components = self._resolve_n_components_for_result(X.shape)
self.n_components_ = n_components
self.noise_variance_ = self._compute_noise_variance(n_components, n_sf_min)

if n_components < params["n_components"]:
self.explained_variance_ = self.explained_variance_[:n_components]
Expand Down
4 changes: 2 additions & 2 deletions onedal/decomposition/tests/test_incremental_pca.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -183,8 +183,8 @@ def test_on_random_data(
if len(sorted_eigenvalues) > n_components
else 0.0
)
# TODO Fix noise variance computation (It is necessary to update C++ side)
# assert np.abs(incpca.noise_variance_ - expected_noise_variance) < tol

assert np.abs(incpca.noise_variance_ - expected_noise_variance) < tol

expected_transformed_data = centered_data @ components.T
if whiten:
Expand Down
4 changes: 2 additions & 2 deletions sklearnex/preview/decomposition/tests/test_incremental_pca.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -167,8 +167,8 @@ def check_pca(incpca, dtype, whiten, data, transformed_data):
if len(sorted_eigenvalues) > n_components
else 0.0
)
# TODO Fix noise variance computation (It is necessary to update C++ side)
# assert np.abs(incpca.noise_variance_ - expected_noise_variance) < tol

assert np.abs(incpca.noise_variance_ - expected_noise_variance) < tol

expected_transformed_data = centered_data @ components.T
if whiten:
Expand Down
Loading