def test_arithmetic_operators(self):
X_3d = hp.Dataset(self.data_3d)
X_4d = hp.Dataset(self.data_4d)
spectral_array = np.random.rand(32)
# Multiplication
X_3d *= 2
X_3d *= spectral_array
X_4d *= 2
X_4d *= spectral_array
# Division
X_3d /= 2
X_3d /= spectral_array
X_4d /= 2
X_4d /= spectral_array
# Addition
X_3d += 2
X_3d += spectral_array
X_4d += 2
X_4d += spectral_array
# Subtraction
X_3d -= 2
X_3d -= spectral_array
X_4d -= 2
X_4d -= spectral_array
def test_preprocessing(self):
X_3d = hp.Dataset(self.data_3d)
X_4d = hp.Dataset(self.data_4d)
for preprocess_type in PREPROCESSING_TYPES:
X_3d.preprocess(mdl=preprocess_type())
X_4d.preprocess(mdl=preprocess_type())
def test_scree(self):
X_3d = hp.Dataset(self.data_3d)
X_4d = hp.Dataset(self.data_4d)
scree_3d = X_3d.scree()
scree_4d = X_4d.scree()
assert scree_3d.shape == (X_3d.shape[-1], )
assert scree_4d.shape == (X_4d.shape[-1], )
def test_data_types(self):
data_empty_list = []
data_1d = np.random.rand(20)
data_2d = np.random.rand(2, 20)
# data type checks
with pytest.raises(TypeError): hp.Dataset(data_empty_list)
with pytest.raises(TypeError): hp.Dataset(data_1d)
with pytest.raises(TypeError): hp.Dataset(data_2d)
def test_flatten(self):
X_3d = hp.Dataset(self.data_3d)
X_4d = hp.Dataset(self.data_4d)
flattened_3d = X_3d.flatten()
flattened_4d = X_4d.flatten()
assert flattened_3d.shape == (X_3d.shape[0] * X_3d.shape[1],
X_3d.shape[2])
assert flattened_4d.shape == (X_4d.shape[0] * X_4d.shape[1] *
X_4d.shape[2], X_4d.shape[3])
def test_cluster(self):
X_3d = hp.Dataset(self.data_3d)
X_4d = hp.Dataset(self.data_4d)
for cluster_type in CLUSTER_TYPES:
if not type(cluster_type()) in (AffinityPropagation, DBSCAN, MeanShift):
lbls_3d, spcs3d = X_3d.cluster(
mdl=cluster_type(n_clusters=2),
decomposed=False
)
lbls_4d, spcs4d = X_4d.cluster(
mdl=cluster_type(n_clusters=2),
decomposed=False
)
lbls_decomp_3d, spcs_decomp_3d = X_3d.cluster(
mdl=cluster_type(n_clusters=2),
decomposed=True,
pca_comps=2
)
lbls_decomp_4d, spcs_decomp_4d = X_4d.cluster(
mdl=cluster_type(n_clusters=2),
decomposed=True,
pca_comps=2
)
assert lbls_3d.shape == (8, 8)
assert lbls_4d.shape == (8, 8, 2)
assert spcs3d.shape == (32, 2)
assert spcs4d.shape == (32, 2)
assert lbls_decomp_3d.shape == (8, 8)
assert lbls_decomp_4d.shape == (8, 8, 2)
assert spcs_decomp_3d.shape == (32, 2)
assert spcs_decomp_4d.shape == (32, 2)
elif type(cluster_type()) in (AffinityPropagation, DBSCAN, MeanShift):
X_3d.cluster(
mdl=cluster_type(),
decomposed=False
)
X_4d.cluster(
mdl=cluster_type(),
decomposed=False
)
X_3d.cluster(
mdl=cluster_type(),
decomposed=True,
pca_comps=2
)
X_4d.cluster(
mdl=cluster_type(),
decomposed=True,
pca_comps=2
)
def test_decompose(self):
X_3d = hp.Dataset(self.data_3d)
X_4d = hp.Dataset(self.data_4d)
for decomp_type in DECOMPOSE_TYPES:
ims_3d, spcs3d = X_3d.decompose(mdl=decomp_type(n_components=2))
ims_4d, spcs4d = X_4d.decompose(mdl=decomp_type(n_components=2))
assert ims_3d.shape == (8, 8, 2)
assert spcs3d.shape == (32, 2)
assert ims_4d.shape == (8, 8, 2, 2)
assert spcs4d.shape == (32, 2)
def test_data_access(self):
X_3d = hp.Dataset(self.data_3d)
X_4d = hp.Dataset(self.data_4d)
arr3d = X_3d[:2, :2, :]
arr4d = X_4d[:2, :2, :1, :]
assert arr3d.shape == (2, 2, 32)
assert arr4d.shape == (2, 2, 1, 32)
X_3d[0, 0, :] = np.random.rand(32)
X_4d[0, 0, 0, :] = np.random.rand(32)
def test_abundance(self):
X_3d = hp.Dataset(self.data_3d)
X_4d = hp.Dataset(self.data_4d)
lbls_3d = X_3d.abundance(spectra=np.random.rand(32, 5),
plot=False,
return_arrs=True)
lbls_4d = X_4d.abundance(spectra=np.random.rand(32, 5),
plot=False,
return_arrs=True)
assert lbls_3d.shape == (8, 8, 5)
assert lbls_4d.shape == (8, 8, 2, 5)
def test_smoothing(self):
X_3d = hp.Dataset(self.data_3d)
X_3d.smoothing = 'savitzky_golay'
X_3d.smoothen(window_length=5, polyorder=3)
X_3d = hp.Dataset(self.data_3d)
X_3d.smoothing = 'gaussian_filter'
X_3d.smoothen(sigma=0.5)
X_4d = hp.Dataset(self.data_4d)
X_4d.smoothing = 'savitzky_golay'
X_4d.smoothen(window_length=5, polyorder=3)
X_4d = hp.Dataset(self.data_4d)
X_4d.smoothing = 'gaussian_filter'
X_4d.smoothen(sigma=0.5)
def test_mixture(self):
X_3d = hp.Dataset(self.data_3d)
X_4d = hp.Dataset(self.data_4d)
for mixture_type in MIXTURE_TYPES:
lbls_3d, spcs_3d = X_3d.mixture(mdl=mixture_type(n_components=2),
plot=False,
return_arrs=True)
lbls_4d, spcs_4d = X_4d.mixture(mdl=mixture_type(n_components=2),
plot=False,
return_arrs=True)
assert lbls_3d.shape == (8, 8)
assert lbls_4d.shape == (8, 8, 2)
assert spcs_3d.shape == (32, 2)
assert spcs_4d.shape == (32, 2)
def test_vca(self):
X_3d = hp.Dataset(self.data_3d)
X_4d = hp.Dataset(self.data_4d)
spectra_3d, coords_3d = X_3d.vca(
n_components=2,
plot=False,
return_arrs=True
)
spectra_4d, coords_4d = X_4d.vca(
n_components=2,
plot=False,
return_arrs=True
)
assert spectra_3d.shape == (32, 2)
assert spectra_4d.shape == (32, 2)
def test_preprocess_methods(self):
X_3d = hp.Dataset(self.data_3d)
X_3d.scale()
X_3d = hp.Dataset(self.data_3d)
X_3d.robust_scale()
X_3d = hp.Dataset(self.data_3d)
X_3d.normalize()
X_4d = hp.Dataset(self.data_4d)
X_4d.scale()
X_4d = hp.Dataset(self.data_4d)
X_4d.robust_scale()
X_4d = hp.Dataset(self.data_4d)
X_4d.normalize()
def test_for_fail(self):
X_3d = hp.Dataset(self.data_3d)
with pytest.raises(TypeError):
X_3d.preprocess(mdl=np.zeros(2))
def test_print_dataset(self):
X_3d = hp.Dataset(self.data_3d)
X_4d = hp.Dataset(self.data_4d)
print(X_3d)
print(X_4d)