def test_hierarchical_no_hartigans_rule(self, model_national):
data = model_national._model_data
with pytest.raises(exceptions.ModelError) as excinfo:
funcs.apply_clustering(data,
timesteps=None,
clustering_func='hierarchical',
how='mean',
normalize=True)
assert check_error_or_warning(
excinfo, 'Cannot undertake hierarchical clustering')
def test_hartigans_rule(self, model_national):
data = model_national._model_data
with pytest.warns(exceptions.ModelWarning) as excinfo:
funcs.apply_clustering(data,
timesteps=None,
clustering_func='kmeans',
how='mean',
normalize=True)
assert check_error_or_warning(excinfo,
'a good number of clusters is 5')
def test_hierarchical_no_hartigans_rule(self, model_national):
data = model_national._model_data
with pytest.raises(exceptions.ModelError) as excinfo:
funcs.apply_clustering(
data,
timesteps=None,
clustering_func='hierarchical',
how='mean',
normalize=True
)
assert check_error_or_warning(excinfo, 'Cannot undertake hierarchical clustering')
def test_hartigans_rule(self, model_national):
data = model_national._model_data
with pytest.warns(exceptions.ModelWarning) as excinfo:
funcs.apply_clustering(
data,
timesteps=None,
clustering_func='kmeans',
how='mean',
normalize=True
)
assert check_error_or_warning(excinfo, 'a good number of clusters is 5')
def test_hartigans_rule(self, model_national):
data = model_national._model_data
with pytest.warns(exceptions.ModelWarning) as excinfo:
funcs.apply_clustering(data,
timesteps=None,
clustering_func='get_clusters_kmeans',
how='mean',
normalize=True)
all_warnings = ','.join(
str(excinfo.list[i]) for i in range(len(excinfo.list)))
assert '5 is a good number of clusters' in all_warnings
def test_hierarchical_closest(self, model_national):
data = model_national._model_data
data_clustered = funcs.apply_clustering(data,
timesteps=None,
clustering_func='hierarchical',
how='closest',
normalize=True,
k=5)
def test_15min_to_2h_clustering(self):
# The data is identical for '2005-01-01' and '2005-01-03' timesteps,
# it is only different for '2005-01-02'
override = {
'techs.test_demand_elec.constraints.resource': 'file=demand_elec_15T_to_2h.csv',
'model.subset_time': None
}
model = build_test_model(override, scenario='simple_supply,one_day')
data = model._model_data
data_clustered_kmeans = funcs.apply_clustering(
data,
timesteps=None,
clustering_func='kmeans',
how='mean',
normalize=True,
k=2
)
data_clustered_hierarchical = funcs.apply_clustering(
data,
timesteps=None,
clustering_func='hierarchical',
how='mean',
normalize=True,
k=2
)
assert len(data_clustered_kmeans.clusters.to_pandas().unique()) == 2
assert len(data_clustered_hierarchical.clusters.to_pandas().unique()) == 2
days = np.unique(data_clustered_kmeans.timesteps.to_index().strftime('%Y-%m-%d'))
# not sure which of '2005-01-01' and '2005-01-03' it will choose to
# label the cluster of those two days
assert '2005-01-02' in days and ('2005-01-01' in days or '2005-01-03' in days)
assert np.array_equal(
data_clustered_kmeans.timestep_resolution.values,
[
0.25, 0.25, 0.25, 0.25, 0.5, 0.5, 0.5, 0.5, 1, 1, 1, 2, 2, 2, 2,
2, 1, 1, 2, 2, 2, 0.25, 0.25, 0.25, 0.25, 0.5, 0.5, 0.5, 0.5, 1,
1, 1, 2, 2, 2, 2, 2, 1, 1, 2, 2, 2
]
)
def test_kmeans_closest(self, model_national):
data = model_national._model_data
data_clustered = funcs.apply_clustering(
data,
timesteps=None,
clustering_func='get_clusters_kmeans',
how='closest',
normalize=True,
k=5)
def test_hierarchical_closest(self, model_national):
data = model_national._model_data
data_clustered = funcs.apply_clustering(
data,
timesteps=None,
clustering_func='hierarchical',
how='closest',
normalize=True,
k=5
)
def test_hierarchical_mean(self, model_national):
data = model_national._model_data
data_clustered = funcs.apply_clustering(data,
timesteps=None,
clustering_func='hierarchical',
how='mean',
normalize=True,
k=5)
assert len(data_clustered.clusters.to_pandas().unique()) == 5
def test_hierarchical_mean(self, model_national):
data = model_national._model_data
data_clustered = funcs.apply_clustering(
data,
timesteps=None,
clustering_func='hierarchical',
how='mean',
normalize=True,
k=5
)
assert len(data_clustered.clusters.to_pandas().unique()) == 5
