Skip to content

TST use global_dtype in sklearn/cluster/tests/test_optics.py #22668

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

Merged
merged 4 commits into from
Apr 1, 2022
Merged

TST use global_dtype in sklearn/cluster/tests/test_optics.py #22668

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
38a6fb9
TST Adapt test_optics.py to test implementations on 32bit datasets
jjerphan Mar 3, 2022
2d2ed3b
Merge branch 'main' into tst/test_optics-32bit
jjerphan Mar 24, 2022
f63175e
TST Use global_dtype
jjerphan Mar 24, 2022
0cd111e
Apply review comments
jjerphan Mar 30, 2022
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
42 changes: 26 additions & 16 deletions sklearn/cluster/tests/test_optics.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -83,7 +83,7 @@ def test_the_extract_xi_labels(ordering, clusters, expected):
assert_array_equal(labels, expected)


def test_extract_xi():
def test_extract_xi(global_dtype):
# small and easy test (no clusters around other clusters)
# but with a clear noise data.
rng = np.random.RandomState(0)
Expand All @@ -96,7 +96,9 @@ def test_extract_xi():
C5 = [3, -2] + 0.6 * rng.randn(n_points_per_cluster, 2)
C6 = [5, 6] + 0.2 * rng.randn(n_points_per_cluster, 2)

X = np.vstack((C1, C2, C3, C4, C5, np.array([[100, 100]]), C6))
X = np.vstack((C1, C2, C3, C4, C5, np.array([[100, 100]]), C6)).astype(
global_dtype, copy=False
)
expected_labels = np.r_[[2] * 5, [0] * 5, [1] * 5, [3] * 5, [1] * 5, -1, [4] * 5]
X, expected_labels = shuffle(X, expected_labels, random_state=rng)

Expand All @@ -111,7 +113,9 @@ def test_extract_xi():
).fit(X)
assert_array_equal(clust.labels_, expected_labels)

X = np.vstack((C1, C2, C3, C4, C5, np.array([[100, 100]] * 2), C6))
X = np.vstack((C1, C2, C3, C4, C5, np.array([[100, 100]] * 2), C6)).astype(
global_dtype, copy=False
)
expected_labels = np.r_[
[1] * 5, [3] * 5, [2] * 5, [0] * 5, [2] * 5, -1, -1, [4] * 5
]
Expand All @@ -126,7 +130,7 @@ def test_extract_xi():
C1 = [[0, 0], [0, 0.1], [0, -0.1], [0.1, 0]]
C2 = [[10, 10], [10, 9], [10, 11], [9, 10]]
C3 = [[100, 100], [100, 90], [100, 110], [90, 100]]
X = np.vstack((C1, C2, C3))
X = np.vstack((C1, C2, C3)).astype(global_dtype, copy=False)
expected_labels = np.r_[[0] * 4, [1] * 4, [2] * 4]
X, expected_labels = shuffle(X, expected_labels, random_state=rng)

Expand All @@ -136,11 +140,15 @@ def test_extract_xi():
assert_array_equal(clust.labels_, expected_labels)


def test_cluster_hierarchy_():
def test_cluster_hierarchy_(global_dtype):
rng = np.random.RandomState(0)
n_points_per_cluster = 100
C1 = [0, 0] + 2 * rng.randn(n_points_per_cluster, 2)
C2 = [0, 0] + 50 * rng.randn(n_points_per_cluster, 2)
C1 = [0, 0] + 2 * rng.randn(n_points_per_cluster, 2).astype(
global_dtype, copy=False
)
C2 = [0, 0] + 50 * rng.randn(n_points_per_cluster, 2).astype(
global_dtype, copy=False
)
X = np.vstack((C1, C2))
X = shuffle(X, random_state=0)

Expand Down Expand Up @@ -278,14 +286,16 @@ def test_close_extract():

@pytest.mark.parametrize("eps", [0.1, 0.3, 0.5])
@pytest.mark.parametrize("min_samples", [3, 10, 20])
def test_dbscan_optics_parity(eps, min_samples):
def test_dbscan_optics_parity(eps, min_samples, global_dtype):
# Test that OPTICS clustering labels are <= 5% difference of DBSCAN

centers = [[1, 1], [-1, -1], [1, -1]]
X, labels_true = make_blobs(
n_samples=750, centers=centers, cluster_std=0.4, random_state=0
)

X = X.astype(global_dtype, copy=False)

# calculate optics with dbscan extract at 0.3 epsilon
op = OPTICS(min_samples=min_samples, cluster_method="dbscan", eps=eps).fit(X)

Expand All @@ -304,11 +314,11 @@ def test_dbscan_optics_parity(eps, min_samples):
assert percent_mismatch <= 0.05


def test_min_samples_edge_case():
def test_min_samples_edge_case(global_dtype):
C1 = [[0, 0], [0, 0.1], [0, -0.1]]
C2 = [[10, 10], [10, 9], [10, 11]]
C3 = [[100, 100], [100, 96], [100, 106]]
X = np.vstack((C1, C2, C3))
X = np.vstack((C1, C2, C3)).astype(global_dtype, copy=False)

expected_labels = np.r_[[0] * 3, [1] * 3, [2] * 3]
clust = OPTICS(min_samples=3, max_eps=7, cluster_method="xi", xi=0.04).fit(X)
Expand All @@ -326,8 +336,8 @@ def test_min_samples_edge_case():

# try arbitrary minimum sizes
@pytest.mark.parametrize("min_cluster_size", range(2, X.shape[0] // 10, 23))
def test_min_cluster_size(min_cluster_size):
redX = X[::2] # reduce for speed
def test_min_cluster_size(min_cluster_size, global_dtype):
redX = X[::2].astype(global_dtype, copy=False) # reduce for speed
clust = OPTICS(min_samples=9, min_cluster_size=min_cluster_size).fit(redX)
cluster_sizes = np.bincount(clust.labels_[clust.labels_ != -1])
if cluster_sizes.size:
Expand Down Expand Up @@ -771,7 +781,7 @@ def test_wrong_cluster_method():
clust.fit(X)


def test_extract_dbscan():
def test_extract_dbscan(global_dtype):
# testing an easy dbscan case. Not including clusters with different
# densities.
rng = np.random.RandomState(0)
Expand All @@ -780,14 +790,14 @@ def test_extract_dbscan():
C2 = [4, -1] + 0.2 * rng.randn(n_points_per_cluster, 2)
C3 = [1, 2] + 0.2 * rng.randn(n_points_per_cluster, 2)
C4 = [-2, 3] + 0.2 * rng.randn(n_points_per_cluster, 2)
X = np.vstack((C1, C2, C3, C4))
X = np.vstack((C1, C2, C3, C4)).astype(global_dtype, copy=False)

clust = OPTICS(cluster_method="dbscan", eps=0.5).fit(X)
assert_array_equal(np.sort(np.unique(clust.labels_)), [0, 1, 2, 3])


def test_precomputed_dists():
redX = X[::2]
def test_precomputed_dists(global_dtype):
redX = X[::2].astype(global_dtype, copy=False)
dists = pairwise_distances(redX, metric="euclidean")
clust1 = OPTICS(min_samples=10, algorithm="brute", metric="precomputed").fit(dists)
clust2 = OPTICS(min_samples=10, algorithm="brute", metric="euclidean").fit(redX)
Expand Down