Skip to content
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

Jump to bottom

[MRG] Make OPTICS support sparse input #14736

Closed
wants to merge 6 commits into from
Closed

[MRG] Make OPTICS support sparse input #14736

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
5f7e6de
Try csr support.
huntzhan Aug 23, 2019
18425aa
Change the default metric of OPTICS to euclidean.
huntzhan Aug 23, 2019
f712b46
Retain the default metric minkowski.
huntzhan Aug 26, 2019
876bb29
Undo tests.
huntzhan Aug 26, 2019
71b5c1e
Fix flake8.
huntzhan Aug 26, 2019
6f498a9
Add sparse tests.
huntzhan Aug 26, 2019
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Jump to
Jump to file
Failed to load files.
Diff view
Diff view
6 changes: 3 additions & 3 deletions sklearn/cluster/optics_.py
View file
Open in desktop
Expand Up @@ -194,7 +194,7 @@ class OPTICS(BaseEstimator, ClusterMixin):
the Conference "Lernen, Wissen, Daten, Analysen" (LWDA) (2018): 318-329.
"""

def __init__(self, min_samples=5, max_eps=np.inf, metric='minkowski', p=2,
def __init__(self, min_samples=5, max_eps=np.inf, metric='euclidean', p=2,
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The default metric minkowski doesn't accept the sparse matrix.

See https://github.com/huntzhan/scikit-learn/blob/5f7e6de69e425cfdfeeb82cf692b81fe22589dd5/sklearn/metrics/pairwise.py#L1546-L1555

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

But this won't work if the user changes p. So I'd rather reinterpret the metric as euclidean below when using the metric, rather than changing the default

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

make sense.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Are you able to add commits to resolve it?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yeah, I will resolve this issue and add test cases this week.

metric_params=None, cluster_method='xi', eps=None, xi=0.05,
predecessor_correction=True, min_cluster_size=None,
algorithm='auto', leaf_size=30, n_jobs=None):
Expand Down Expand Up @@ -233,7 +233,7 @@ def fit(self, X, y=None):
self : instance of OPTICS
The instance.
"""
X = check_array(X, dtype=np.float)
X = check_array(X, accept_sparse='csr')

if self.cluster_method not in ['dbscan', 'xi']:
raise ValueError("cluster_method should be one of"
Expand Down Expand Up @@ -517,7 +517,7 @@ def _set_reach_dist(core_distances_, reachability_, predecessor_,
# the same logic as neighbors, p is ignored if explicitly set
# in the dict params
_params['p'] = p
dists = pairwise_distances(P, np.take(X, unproc, axis=0),
dists = pairwise_distances(P, X[unproc],
Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

np.take doesn't support the sparse matrix

metric, n_jobs=None,
**_params).ravel()

Expand Down
4 changes: 2 additions & 2 deletions sklearn/cluster/tests/test_optics.py
View file
Open in desktop
Expand Up @@ -352,7 +352,7 @@ def test_compare_to_ELKI():
# Tests against known extraction array
# Does NOT work with metric='euclidean', because sklearn euclidean has
# worse numeric precision. 'minkowski' is slower but more accurate.
clust1 = OPTICS(min_samples=5).fit(X)
clust1 = OPTICS(metric='minkowski', min_samples=5).fit(X)

assert_array_equal(clust1.ordering_, np.array(o1))
assert_array_equal(clust1.predecessor_[clust1.ordering_], np.array(p1))
Expand Down Expand Up @@ -386,7 +386,7 @@ def test_compare_to_ELKI():
11, 19, 15, 10, 47, -1, 20, 22, 25, 25, 25, 25, 22, 22, 23, -1, 30,
30, 34, 34, 34, 32, 32, 37, 38, -1, -1, -1, -1, -1, -1, -1, -1, -1,
-1, -1, -1, -1, -1, -1, -1, -1, -1]
clust2 = OPTICS(min_samples=5, max_eps=0.5).fit(X)
clust2 = OPTICS(metric='minkowski', min_samples=5, max_eps=0.5).fit(X)

assert_array_equal(clust2.ordering_, np.array(o2))
assert_array_equal(clust2.predecessor_[clust2.ordering_], np.array(p2))
Expand Down