def test_persistence_lead0_lead1(hind_ds_initialized_1d,
hind_ds_initialized_1d_lead0,
reconstruction_ds_1d):
"""
Checks that compute persistence returns the same results with a lead-0 and lead-1
framework.
"""
res1 = compute_persistence(hind_ds_initialized_1d,
reconstruction_ds_1d,
metric='rmse')
res2 = compute_persistence(hind_ds_initialized_1d_lead0,
reconstruction_ds_1d,
metric='rmse')
assert (res1.SST.values == res2.SST.values).all()
def test_compute_persistence_ds1d_not_nan(
PM_ds_initialized_1d, PM_ds_control_1d, metric
):
"""
Checks that there are no NaNs on persistence forecast of 1D time series.
"""
if metric == "contingency":
metric_kwargs = {
"forecast_category_edges": category_edges,
"observation_category_edges": category_edges,
"score": "accuracy",
}
else:
metric_kwargs = {}
actual = (
compute_persistence(
PM_ds_initialized_1d,
PM_ds_control_1d,
metric=metric,
alignment="same_inits",
**metric_kwargs
)
# .isnull()
# .any()
)
print(actual.tos)
actual = actual.isnull().any()
for var in actual.data_vars:
assert not actual[var], actual[var]
def test_compute_persistence_lead0_lead1(
PM_da_initialized_1d, PM_da_initialized_1d_lead0, PM_da_control_1d, metric
):
"""
Checks that persistence forecast results are identical for a lead 0 and lead 1 setup
"""
res1 = compute_persistence(
PM_da_initialized_1d, PM_da_control_1d, metric=metric, alignment="same_inits"
)
res2 = compute_persistence(
PM_da_initialized_1d_lead0,
PM_da_control_1d,
metric=metric,
alignment="same_inits",
)
assert (res1.values == res2.values).all()
def test_persistence(hind_da_initialized_1d, reconstruction_da_1d, metric):
"""
Checks that compute persistence works without breaking.
"""
res = compute_persistence(hind_da_initialized_1d,
reconstruction_da_1d,
metric=metric)
assert not res.isnull().any()
# check persistence metadata
assert res.attrs['metric'] == metric
assert res.attrs['skill_calculated_by_function'] == 'compute_persistence'
assert 'number of members' not in res.attrs
def test_compute_persistence_add_attrs(PM_ds_initialized_1d, PM_ds_control_1d,
metric):
"""
Checks that there are no NaNs on persistence forecast of 1D time series.
"""
attrs = (compute_persistence(
PM_ds_initialized_1d,
PM_ds_control_1d,
metric=metric,
alignment="same_inits",
)).attrs
assert (attrs["prediction_skill"] ==
"calculated by climpred https://climpred.readthedocs.io/")
assert attrs["skill_calculated_by_function"] == "compute_persistence"
assert "number of members" not in attrs
assert attrs["metric"] == metric
def test_compute_persistence_ds1d_not_nan(
PM_ds_initialized_1d, PM_ds_control_1d, metric
):
"""
Checks that there are no NaNs on persistence forecast of 1D time series.
"""
actual = (
compute_persistence(
PM_ds_initialized_1d,
PM_ds_control_1d,
metric=metric,
alignment='same_inits',
)
.isnull()
.any()
)
for var in actual.data_vars:
assert not actual[var]
def test_compute_persistence_add_attrs(PM_ds_initialized_1d, PM_ds_control_1d, metric):
"""
Checks that there are no NaNs on persistence forecast of 1D time series.
"""
attrs = (
compute_persistence(
PM_ds_initialized_1d,
PM_ds_control_1d,
metric=metric,
alignment='same_inits',
)
).attrs
assert (
attrs['prediction_skill']
== 'calculated by climpred https://climpred.readthedocs.io/'
)
assert attrs['skill_calculated_by_function'] == 'compute_persistence'
assert 'number of members' not in attrs
assert attrs['metric'] == metric
def test_persistence(hind_da_initialized_1d, reconstruction_da_1d, metric):
"""
Checks that compute persistence works without breaking.
"""
if metric == "contingency":
metric_kwargs = {
"forecast_category_edges": category_edges,
"observation_category_edges": category_edges,
"score": "accuracy",
}
else:
metric_kwargs = {}
res = compute_persistence(
hind_da_initialized_1d, reconstruction_da_1d, metric=metric, **metric_kwargs
)
assert not res.isnull().any()
# check persistence metadata
assert res.attrs["metric"] == metric
assert res.attrs["skill_calculated_by_function"] == "compute_persistence"
assert "number of members" not in res.attrs
