def test_candidate_model_json_with_warning():
eemeter_warning = EEMeterWarning(qualified_name="qualified_name",
description="description",
data={})
candidate_model = CandidateModel(
model_type="model_type",
formula="formula",
status="status",
warnings=[eemeter_warning],
)
assert candidate_model.json() == {
"formula":
"formula",
"model_params": {},
"model_type":
"model_type",
"r_squared_adj":
None,
"status":
"status",
"warnings": [{
"data": {},
"description": "description",
"qualified_name": "qualified_name",
}],
}
def test_plot_caltrack_candidate_error():
candidate_model = CandidateModel(
model_type="intercept_only",
formula="formula",
status="ERROR",
plot_func=plot_caltrack_candidate, # no predict func
model_params={"intercept": 1},
)
ax = candidate_model.plot()
assert ax is None
def test_plot_caltrack_candidate_best():
candidate_model = CandidateModel(
model_type="intercept_only",
formula="formula",
status="QUALIFIED",
predict_func=caltrack_predict,
plot_func=plot_caltrack_candidate,
model_params={"intercept": 1},
)
ax = candidate_model.plot(best=True)
data = ax.lines[0].get_xydata()
assert data.shape == (60, 2)
def test_plot_caltrack_candidate_qualified():
candidate_model = CandidateModel(
model_type="intercept_only",
formula="formula",
status="QUALIFIED",
predict_func=caltrack_predict,
plot_func=plot_caltrack_candidate,
model_params={"intercept": 1},
)
ax = candidate_model.plot(candidate_model, title="title")
data = ax.lines[0].get_xydata()
assert data.shape == (60, 2)
assert ax.get_title() == "title"
def test_candidate_model_with_predict_func():
def predict_func(model_type, model_params, input_):
return input_
candidate_model = CandidateModel(
model_type="model_type",
formula="formula",
status="status",
predict_func=predict_func,
)
assert candidate_model.model_type == "model_type"
assert candidate_model.formula == "formula"
assert candidate_model.status == "status"
assert candidate_model.predict("a") == "a"
def test_model_results_json_with_model_metrics():
candidate_model = CandidateModel(model_type="model_type",
formula="formula",
status="status",
r_squared_adj=0.5)
model_results = ModelResults(
status="status",
method_name="method_name",
model=candidate_model,
r_squared_adj=np.nan,
)
model_metrics = ModelMetrics(observed_input=pd.Series([0, 1, 2]),
predicted_input=pd.Series([1, 0, 2]))
json_result = model_results.json()
json.dumps(json_result) # just make sure it's valid json
json_result["metrics"] = {} # overwrite because of floats
assert json_result == {
"candidates": None,
"metadata": {},
"method_name": "method_name",
"metrics": {},
"model": {
"formula": "formula",
"model_params": {},
"model_type": "model_type",
"r_squared_adj": 0.5,
"status": "status",
"warnings": [],
},
"settings": {},
"status": "status",
"r_squared_adj": None,
"warnings": [],
}
def test_model_results_json_with_nan_r_squared_adj():
candidate_model = CandidateModel(
model_type="model_type",
formula="formula",
status="status",
r_squared_adj=np.nan,
)
model_results = ModelResults(
status="status",
method_name="method_name",
model=candidate_model,
r_squared_adj=np.nan,
)
assert model_results.json() == {
"candidates": None,
"metadata": {},
"method_name": "method_name",
"metrics": None,
"model": {
"formula": "formula",
"model_params": {},
"model_type": "model_type",
"r_squared_adj": None,
"status": "status",
"warnings": [],
},
"settings": {},
"status": "status",
"r_squared_adj": None,
"warnings": [],
}
def test_plot_caltrack_candidate_cdd_hdd_model():
candidate_model = CandidateModel(
model_type="cdd_hdd",
formula="formula",
status="QUALIFIED",
predict_func=caltrack_predict,
plot_func=plot_caltrack_candidate,
model_params={
"beta_hdd": 1,
"beta_cdd": 1,
"cooling_balance_point": 65,
"heating_balance_point": 65,
"intercept": 1,
},
)
ax = candidate_model.plot()
data = ax.lines[0].get_xydata()
assert data.shape == (60, 2)
def test_candidate_model_minimal():
candidate_model = CandidateModel(model_type="model_type",
formula="formula",
status="status")
assert candidate_model.model_type == "model_type"
assert candidate_model.formula == "formula"
assert candidate_model.status == "status"
with pytest.raises(ValueError):
candidate_model.predict("a") == "a"
assert candidate_model.model_params == {}
assert candidate_model.warnings == []
assert str(candidate_model).startswith("CandidateModel")
assert candidate_model.json() == {
"formula": "formula",
"model_params": {},
"model_type": "model_type",
"r_squared_adj": None,
"status": "status",
"warnings": [],
}
def test_model_results_json_with_objects():
candidate_model = CandidateModel(model_type="model_type",
formula="formula",
status="status")
eemeter_warning = EEMeterWarning(qualified_name="qualified_name",
description="description",
data={})
model_results = ModelResults(
status="status",
method_name="method_name",
model=candidate_model,
candidates=[candidate_model],
warnings=[eemeter_warning],
)
assert model_results.json(with_candidates=True) == {
"candidates": [{
"formula": "formula",
"model_params": {},
"model_type": "model_type",
"r_squared_adj": None,
"status": "status",
"warnings": [],
}],
"metadata": {},
"method_name":
"method_name",
"metrics":
None,
"model": {
"formula": "formula",
"model_params": {},
"model_type": "model_type",
"r_squared_adj": None,
"status": "status",
"warnings": [],
},
"settings": {},
"status":
"status",
"r_squared_adj":
None,
"warnings": [{
"data": {},
"description": "description",
"qualified_name": "qualified_name",
}],
}
def test_plot_model_results(il_electricity_cdd_hdd_daily):
candidate_model = CandidateModel(
model_type="intercept_only",
formula="formula",
status="QUALIFIED",
predict_func=caltrack_predict,
plot_func=plot_caltrack_candidate,
model_params={"intercept": 1},
)
model_results = ModelResults(
status="status",
method_name="method_name",
model=candidate_model,
candidates=[candidate_model],
)
ax = model_results.plot(title="title", with_candidates=True)
data = ax.lines[0].get_xydata()
assert data.shape == (70, 2)
data = ax.lines[1].get_xydata()
assert data.shape == (70, 2)
assert ax.get_title() == "title"
def test_candidate_model_with_no_plot_func():
candidate_model = CandidateModel(model_type="model_type",
formula="formula",
status="status")
with pytest.raises(ValueError):
candidate_model.plot("a")