def test_report(self, model):
reg_rep = model.report()
assert reg_rep["index"] == [
"explained_variance",
"max_error",
"median_absolute_error",
"mean_absolute_error",
"mean_squared_error",
"root_mean_squared_error",
"r2",
"r2_adj",
"aic",
"bic",
]
assert reg_rep["value"][0] == pytest.approx(0.219816, abs=1e-6)
assert reg_rep["value"][1] == pytest.approx(3.592465, abs=1e-6)
assert reg_rep["value"][2] == pytest.approx(0.496031, abs=1e-6)
assert reg_rep["value"][3] == pytest.approx(0.609075, abs=1e-6)
assert reg_rep["value"][4] == pytest.approx(0.594856, abs=1e-6)
assert reg_rep["value"][5] == pytest.approx(0.7712695123858948,
abs=1e-6)
assert reg_rep["value"][6] == pytest.approx(0.219816, abs=1e-6)
assert reg_rep["value"][7] == pytest.approx(0.21945605202370688,
abs=1e-6)
assert reg_rep["value"][8] == pytest.approx(-3366.7617912479104,
abs=1e-6)
assert reg_rep["value"][9] == pytest.approx(-3339.65156943384,
abs=1e-6)
model_class = Pipeline([
(
"NormalizerWine",
StandardScaler("logstd_model_test", cursor=model.cursor),
),
(
"LogisticRegressionWine",
LogisticRegression("logreg_model_test", cursor=model.cursor),
),
])
model_class.drop()
model_class.fit("public.winequality", ["alcohol"], "good")
cls_rep1 = model_class.report().transpose()
assert cls_rep1["auc"][0] == pytest.approx(0.7642901826299067)
assert cls_rep1["prc_auc"][0] == pytest.approx(0.45326090911518313)
assert cls_rep1["accuracy"][0] == pytest.approx(0.8131445282438048)
assert cls_rep1["log_loss"][0] == pytest.approx(0.182720882885624)
assert cls_rep1["precision"][0] == pytest.approx(0.5595463137996219)
assert cls_rep1["recall"][0] == pytest.approx(0.2317932654659358)
assert cls_rep1["f1_score"][0] == pytest.approx(0.37307094353346476)
assert cls_rep1["mcc"][0] == pytest.approx(0.2719537880298097)
assert cls_rep1["informedness"][0] == pytest.approx(
0.18715725014026519)
assert cls_rep1["markedness"][0] == pytest.approx(0.3951696381964047)
assert cls_rep1["csi"][0] == pytest.approx(0.19602649006622516)
assert cls_rep1["cutoff"][0] == pytest.approx(0.5)
model_class.drop()
def model(winequality_vd):
model_class = Pipeline([
("NormalizerWine", StandardScaler("std_model_test", )),
("LinearRegressionWine", LinearRegression("linreg_model_test", )),
])
model_class.drop()
model_class.fit("public.winequality",
["citric_acid", "residual_sugar", "alcohol"], "quality")
yield model_class
model_class.drop()
def test_set_cursor(self, base):
model_test = Pipeline([
(
"NormalizerWine",
StandardScaler("std_model_test_vdf", cursor=base.cursor),
),
(
"LinearRegressionWine",
LinearRegression("linreg_model_test_vdf", cursor=base.cursor),
),
])
model_test.drop()
model_test.set_cursor(base.cursor)
model_test.fit("public.winequality", ["alcohol"], "quality")
model_test.cursor.execute(
"SELECT model_name FROM models WHERE model_name IN ('std_model_test_vdf', 'linreg_model_test_vdf')"
)
assert len(base.cursor.fetchall()) == 2
model_test.drop()
def test_transform(self, winequality_vd, model):
model_class = Pipeline([
(
"NormalizerWine",
StandardScaler("logstd_model_test"),
),
(
"NormalizerWine",
MinMaxScaler("logmm_model_test"),
),
])
model_class.drop()
model_class.fit("public.winequality", ["alcohol"])
winequality_copy = winequality_vd.copy()
winequality_copy = model_class.transform(winequality_copy,
X=["alcohol"])
assert winequality_copy["alcohol"].mean() == pytest.approx(
0.361130555239542, abs=1e-6)
model_class.drop()
def test_inverse_transform(self, winequality_vd, model):
model_class = Pipeline([
(
"NormalizerWine",
StandardScaler("logstd_model_test", cursor=model.cursor),
),
(
"NormalizerWine",
MinMaxScaler("logmm_model_test", cursor=model.cursor),
),
])
model_class.drop()
model_class.fit("public.winequality", ["alcohol"])
winequality_copy = winequality_vd.copy()
winequality_copy = model_class.inverse_transform(
winequality_copy,
X=["alcohol"],
)
assert winequality_copy["alcohol"].mean() == pytest.approx(
80.3934257349546, abs=1e-6)
model_class.drop()
def test_model_from_vDF(self, base, winequality_vd):
model_test = Pipeline([
(
"NormalizerWine",
StandardScaler("std_model_test_vdf", cursor=base.cursor),
),
(
"LinearRegressionWine",
LinearRegression("linreg_model_test_vdf", cursor=base.cursor),
),
])
model_test.drop()
model_test.fit(winequality_vd,
["citric_acid", "residual_sugar", "alcohol"], "quality")
model_test.cursor.execute(
"SELECT model_name FROM models WHERE model_name IN ('std_model_test_vdf', 'linreg_model_test_vdf')"
)
assert len(base.cursor.fetchall()) == 2
model_test.drop()
def test_drop(self, base, winequality_vd):
model_class = Pipeline([
(
"NormalizerWine",
StandardScaler("std_model_test_drop", cursor=base.cursor),
),
(
"LinearRegressionWine",
LinearRegression("linreg_model_test_drop", cursor=base.cursor),
),
])
model_class.drop()
model_class.fit(winequality_vd, ["alcohol"], "quality")
model_class.cursor.execute(
"SELECT model_name FROM models WHERE model_name IN ('linreg_model_test_drop', 'std_model_test_drop')"
)
assert len(model_class.cursor.fetchall()) == 2
model_class.drop()
model_class.cursor.execute(
"SELECT model_name FROM models WHERE model_name IN ('linreg_model_test_drop', 'std_model_test_drop')"
)
assert model_class.cursor.fetchone() is None