def test_parameter_df_upsampling_multiple_columns(model):
""" Test that the `DataFrameParameter` works with multiple columns that map to a `Scenario`
"""
scA = Scenario(model, 'A', size=20)
scB = Scenario(model, 'B', size=2)
# scenario indices (not used for this test)
# Use a 7 day timestep for this test and run 2015
model.timestepper.delta = datetime.timedelta(7)
model.timestepper.start = pd.to_datetime('2015-01-01')
model.timestepper.end = pd.to_datetime('2015-12-31')
# Daily time-step
index = pd.date_range('2015-01-01', periods=365, freq='D')
df = pd.DataFrame(np.random.rand(365, 20), index=index)
p = DataFrameParameter(model, df, scenario=scA)
p.setup()
A = df.resample('7D', axis=0).mean()
for v, ts in zip(A.values, model.timestepper):
np.testing.assert_allclose([
p.value(ts, ScenarioIndex(i, np.array([i], dtype=np.int32)))
for i in range(20)
], v)
p = DataFrameParameter(model, df, scenario=scB)
with pytest.raises(ValueError):
p.setup()
def test_timeseries_excel(simple_linear_model, filename):
"""Test creation of a DataFrameParameter from external data (e.g. CSV)"""
model = simple_linear_model
# create DataFrameParameter from external data
filename = os.path.join(TEST_FOLDER, filename)
data = {"url": filename, "column": "Data", "index_col": "Timestamp"}
if filename.endswith(".csv"):
data.update({"parse_dates": True, "dayfirst": True})
ts = DataFrameParameter.load(model, data)
# model (intentionally not aligned)
model.timestepper.start = ts.dataframe.index[0] + 5
model.timestepper.end = ts.dataframe.index[-1] - 12
# need to assign parameter for it's setup method to be called
model.nodes["Input"].max_flow = ts
@assert_rec(model, ts)
def expected(timestep, scenario_index):
return ts.dataframe.loc[timestep.datetime]
model.run()
def test_timeseries_excel(simple_linear_model, filename):
"""Test creation of a DataFrameParameter from external data (e.g. CSV)"""
model = simple_linear_model
# create DataFrameParameter from external data
filename = os.path.join(TEST_FOLDER, filename)
data = {"url": filename, "column": "Data", "index_col": "Timestamp"}
if filename.endswith(".csv"):
data.update({"parse_dates": True, "dayfirst":True})
ts = DataFrameParameter.load(model, data)
# model (intentionally not aligned)
model.timestepper.start = ts.df.index[0] + 5
model.timestepper.end = ts.df.index[-1] - 12
# need to assign parameter for it's setup method to be called
model.nodes["Input"].max_flow = ts
model.setup()
# test accessing a specific value
timestep = build_timestep(model, '2015-01-15')
scenario_index = ScenarioIndex(0, np.array([], dtype=np.int32))
assert(ts.value(timestep, scenario_index) == 28.24)
def test_parameter_df_upsampling(model):
""" Test that the `DataFrameParameter` can upsample data from a `pandas.DataFrame` and return that correctly
"""
# scenario indices (not used for this test)
si = ScenarioIndex(0, np.array([0], dtype=np.int32))
# Use a 7 day timestep for this test and run 2015
model.timestepper.delta = datetime.timedelta(7)
model.timestepper.start = pd.to_datetime('2015-01-01')
model.timestepper.end = pd.to_datetime('2015-12-31')
# Daily time-step
index = pd.date_range('2015-01-01', periods=365, freq='D')
series = pd.Series(np.arange(365), index=index)
p = DataFrameParameter(model, series)
p.setup()
A = series.resample('7D').mean()
for v, ts in zip(A, model.timestepper):
np.testing.assert_allclose(p.value(ts, si), v)
model.reset()
# Daily time-step that requires aligning
index = pd.date_range('2014-12-31', periods=366, freq='D')
series = pd.Series(np.arange(366), index=index)
p = DataFrameParameter(model, series)
p.setup()
# offset the resample appropriately for the test
A = series[1:].resample('7D').mean()
for v, ts in zip(A, model.timestepper):
np.testing.assert_allclose(p.value(ts, si), v)
model.reset()
# Daily time-step that is not covering the require range
index = pd.date_range('2015-02-01', periods=365, freq='D')
series = pd.Series(np.arange(365), index=index)
p = DataFrameParameter(model, series)
with pytest.raises(ValueError):
p.setup()
model.reset()
# Daily time-step that is not covering the require range
index = pd.date_range('2014-11-01', periods=365, freq='D')
series = pd.Series(np.arange(365), index=index)
p = DataFrameParameter(model, series)
with pytest.raises(ValueError):
p.setup()