def test_clustering_is_same_as_scipy():
''' Basic clustering returns the same as scipy '''
features = random_feat_matrix(600, 705)
scipy_ward = sci_hie.ward(features)
our_ward = our_hie.clustering(features)
scipy_centroid = sci_hie.centroid(features)
our_centroid = our_hie.clustering(features, method='centroid')
numpy.testing.assert_almost_equal(scipy_ward, our_ward)
numpy.testing.assert_almost_equal(scipy_centroid, our_centroid)
return True
def test_clustering_is_same_as_scipy_2():
''' Basic clustering returns the same as scipy 2 '''
cifti = nibabel.load('./logpar/cli/tests/data/test.dconn.nii')
features = cifti.get_data()[0, 0, 0, 0]
scipy_ward = sci_hie.ward(features)
our_ward = our_hie.clustering(features)
scipy_centroid = sci_hie.centroid(features)
our_centroid = our_hie.clustering(features, method='centroid')
numpy.testing.assert_almost_equal(scipy_ward[:, :2], our_ward[:, :2])
numpy.testing.assert_almost_equal(scipy_centroid[:, :2],
our_centroid[:, :2])
return True
def test_all_neighbors_is_same_as_scipy():
''' Clustering without constraints returns the same as scipy '''
features = random_feat_matrix(200, 100)
n = features.shape[0]
all_neighbors = numpy.ones((n, n)) - numpy.eye(n)
scipy_ward = sci_hie.ward(features)
our_ward = our_hie.clustering(features, method='ward',
constraints=all_neighbors)
scipy_centroid = sci_hie.centroid(features)
our_centroid = our_hie.clustering(features, method='centroid',
constraints=all_neighbors)
numpy.testing.assert_almost_equal(scipy_ward, our_ward)
numpy.testing.assert_almost_equal(scipy_centroid, our_centroid)
def test_invalid_cases_fails():
''' Tests different cases of wrong configurations which SHOULD fail
TODO: Improve THIS test'''
features = random_feat_matrix(600, 750)
n = features.shape[0]
wrong_const = numpy.ones(n-1)
right_const = numpy.ones((n,n)) - numpy.eye(n)
# 1. Wrong method - 2. Wrong constraints - 3. Wrong clustering size
failing_functions = [lambda: our_hie.clustering(features, method='none'),
lambda: our_hie.clustering(features,
constraints=wrong_const),
lambda: our_hie.clustering(features,
constraints=right_const,
min_size=-3)]
for i, f in enumerate(failing_functions):
if not try_function(f):
raise Exception('Function number {} did not fail!'.format(i))
def test_copy_flag():
''' Tests copy flag '''
features = random_feat_matrix(200, 100)
features_copy = features.copy()
n = features.shape[0]
all_neighbors = numpy.ones((n, n)) - numpy.eye(n)
s = 10
Z = our_hie.clustering(features, constraints=all_neighbors, min_size=s)
numpy.testing.assert_equal(features, features_copy)
Z = our_hie.clustering(features, constraints=all_neighbors, min_size=s,
copy=False)
failed = False
try:
numpy.testing.assert_equal(features, features_copy)
except AssertionError:
failed = True
assert(failed)
def test_minimum_size():
''' Minimum size is respected '''
features = random_feat_matrix(200, 100)
n = features.shape[0]
all_neighbors = numpy.ones((n, n)) - numpy.eye(n)
s = 25
Z = our_hie.clustering(features, constraints=all_neighbors, min_size=s)
# Take the finest granularity
parcels = sci_hie.fcluster(Z, Z[:, 2].min(), criterion='distance')
# Each resulting parcel could be as big as (2s-2), since in the worst case
# two clusters of size (s-1) would be merged
count = Counter(parcels)
assert(all([c <= 2*s for c in count]))
def test_patches_of_neighbors():
''' If the features are organized as patches then the resulting clustering
has contiguous patches '''
features = random_feat_matrix(200, 100)
n = features.shape[0]
# The features are organized in space as disjoint patches
patches = numpy.zeros((n, n))
s = 10
for i in xrange(0, n, s):
patches[i:i+s, i:i+s] = numpy.ones((s, s)) - numpy.eye(s)
# Cluster using patches constraints
Z = our_hie.clustering(features, constraints=patches, min_size=s)
# Take the finest granularity
parcels = sci_hie.fcluster(Z, Z[:, 2].min(), criterion='distance')
# Check they're contiguous patches
for i in xrange(0, n, s):
assert(all([v == parcels[i] for v in parcels[i:i+s]]))