def test_kkz_standardize(self):
ens = open_dataset(self.nc_file)
data = ens.data.isel(criteria=[1, 3, 5])
sel_std = ensembles.kkz_reduce_ensemble(data, 4, standardize=True)
sel_no = ensembles.kkz_reduce_ensemble(data, 4, standardize=False)
assert sel_std == [23, 10, 19, 14]
assert sel_no == [23, 1, 14, 10]
def test_kkz_change_metric(self):
# This test uses stupid values but is meant to test is kwargs are passed and if dist_method is used.
ens = open_dataset(self.nc_file)
data = ens.data.isel(criteria=[1, 3, 5])
sel_euc = ensembles.kkz_reduce_ensemble(data,
4,
dist_method="euclidean")
sel_mah = ensembles.kkz_reduce_ensemble(data,
4,
dist_method="mahalanobis",
VI=np.arange(24))
assert sel_euc == [23, 10, 19, 14]
assert sel_mah == [5, 3, 4, 0]
def test_standardize_seuclidean(self):
# This test the odd choice of standardizing data for a standardized distance metric
ens = open_dataset(self.nc_file)
data = ens.data
for n in np.arange(1, len(data)):
sel1 = ensembles.kkz_reduce_ensemble(data,
n,
dist_method="seuclidean",
standardize=True)
sel2 = ensembles.kkz_reduce_ensemble(data,
n,
dist_method="seuclidean",
standardize=False)
sel3 = ensembles.kkz_reduce_ensemble(data,
n,
dist_method="euclidean",
standardize=True)
assert sel1 == sel2
assert sel1 == sel3
def test_kkz_simple(self, crit, num_select, expected):
ens = open_dataset(self.nc_file)
data = ens.data.isel(criteria=crit)
selected = ensembles.kkz_reduce_ensemble(data, num_select)
assert selected == expected