def test_vDF_score(self, base, titanic_vd):
from verticapy.learn.linear_model import LogisticRegression
model = LogisticRegression(
name="public.LR_titanic",
cursor=base.cursor,
tol=1e-4,
C=1.0,
max_iter=100,
solver="CGD",
l1_ratio=0.5,
)
model.drop() # dropping the model in case of its existance
model.fit("public.titanic", ["fare", "age"], "survived")
model.predict(titanic_vd, name="survived_pred")
# Computing AUC
auc = titanic_vd.score(y_true="survived",
y_score="survived_pred",
method="auc")
assert auc == pytest.approx(0.697476274)
# Computing MSE
mse = titanic_vd.score(y_true="survived",
y_score="survived_pred",
method="mse")
assert mse == pytest.approx(0.224993557)
# Drawing ROC Curve
roc_res = titanic_vd.score(y_true="survived",
y_score="survived_pred",
method="roc")
assert roc_res["threshold"][3] == 0.003
assert roc_res["false_positive"][3] == 1.0
assert roc_res["true_positive"][3] == 1.0
assert roc_res["threshold"][300] == 0.3
assert roc_res["false_positive"][300] == pytest.approx(0.9900826446)
assert roc_res["true_positive"][300] == pytest.approx(0.9974424552)
assert roc_res["threshold"][900] == 0.9
assert roc_res["false_positive"][900] == pytest.approx(0.01818181818)
assert roc_res["true_positive"][900] == pytest.approx(0.06649616368)
# Drawing PRC Curve
prc_res = titanic_vd.score(y_true="survived",
y_score="survived_pred",
method="prc")
assert prc_res["threshold"][3] == 0.002
assert prc_res["recall"][3] == 1.0
assert prc_res["precision"][3] == pytest.approx(0.3925702811)
assert prc_res["threshold"][300] == 0.299
assert prc_res["recall"][300] == pytest.approx(1.0)
assert prc_res["precision"][300] == pytest.approx(0.3949494949)
assert prc_res["threshold"][900] == 0.899
assert prc_res["recall"][900] == pytest.approx(0.06649616368)
assert prc_res["precision"][900] == pytest.approx(0.7027027027)
# dropping the created model
model.drop()
def test_contour(self, titanic_vd):
model_test = LogisticRegression("model_contour",)
model_test.drop()
model_test.fit(
titanic_vd, ["age", "fare"], "survived",
)
result = model_test.contour()
assert len(result.get_default_bbox_extra_artists()) == 38
model_test.drop()
def test_vDF_score(self, base, titanic_vd):
from verticapy.learn.linear_model import LogisticRegression
model = LogisticRegression(
name="public.LR_titanic",
cursor=base.cursor,
tol=1e-4,
C=1.0,
max_iter=100,
solver="CGD",
penalty="ENet",
l1_ratio=0.5,
)
model.drop() # dropping the model in case of its existance
model.fit("public.titanic", ["fare", "age"], "survived")
model.predict(titanic_vd, name="survived_pred")
# Computing AUC
auc = titanic_vd.score(y_true="survived", y_score="survived_pred", method="auc")
assert auc == pytest.approx(0.7051784997146537)
# Computing MSE
mse = titanic_vd.score(y_true="survived", y_score="survived_pred", method="mse")
assert mse == pytest.approx(0.228082579110535)
# Drawing ROC Curve
roc_res = titanic_vd.score(
y_true="survived", y_score="survived_pred", method="roc", nbins=1000,
)
assert roc_res["threshold"][3] == 0.003
assert roc_res["false_positive"][3] == 1.0
assert roc_res["true_positive"][3] == 1.0
assert roc_res["threshold"][300] == 0.3
assert roc_res["false_positive"][300] == pytest.approx(1.0)
assert roc_res["true_positive"][300] == pytest.approx(1.0)
assert roc_res["threshold"][900] == 0.9
assert roc_res["false_positive"][900] == pytest.approx(0.0148760330578512)
assert roc_res["true_positive"][900] == pytest.approx(0.061381074168798)
# Drawing PRC Curve
prc_res = titanic_vd.score(
y_true="survived", y_score="survived_pred", method="prc", nbins=1000,
)
assert prc_res["threshold"][3] == 0.002
assert prc_res["recall"][3] == 1.0
assert prc_res["precision"][3] == pytest.approx(0.3925702811)
assert prc_res["threshold"][300] == 0.299
assert prc_res["recall"][300] == pytest.approx(1.0)
assert prc_res["precision"][300] == pytest.approx(0.392570281124498)
assert prc_res["threshold"][900] == 0.899
assert prc_res["recall"][900] == pytest.approx(0.061381074168798)
assert prc_res["precision"][900] == pytest.approx(0.727272727272727)
# dropping the created model
model.drop()
def test_model_from_vDF(self, base, titanic_vd):
base.cursor.execute("DROP MODEL IF EXISTS logreg_from_vDF")
model_test = LogisticRegression("logreg_from_vDF", cursor=base.cursor)
model_test.fit(titanic_vd, ["age", "fare"], "survived")
base.cursor.execute(
"SELECT model_name FROM models WHERE model_name = 'logreg_from_vDF'"
)
assert base.cursor.fetchone()[0] == "logreg_from_vDF"
model_test.drop()
def test_get_plot(self, base, winequality_vd):
# 1D
base.cursor.execute("DROP MODEL IF EXISTS model_test_plot")
model_test = LogisticRegression("model_test_plot", cursor=base.cursor)
model_test.fit(winequality_vd, ["alcohol"], "good")
result = model_test.plot(color="r")
assert len(result.get_default_bbox_extra_artists()) == 11
plt.close("all")
model_test.drop()
# 2D
model_test.fit(winequality_vd, ["alcohol", "residual_sugar"], "good")
result = model_test.plot(color="r")
assert len(result.get_default_bbox_extra_artists()) == 5
plt.close("all")
model_test.drop()
def test_drop(self, base):
base.cursor.execute("DROP MODEL IF EXISTS logreg_model_test_drop")
model_test = LogisticRegression("logreg_model_test_drop", cursor=base.cursor)
model_test.fit("public.titanic", ["age", "fare"], "survived")
base.cursor.execute(
"SELECT model_name FROM models WHERE model_name = 'logreg_model_test_drop'"
)
assert base.cursor.fetchone()[0] == "logreg_model_test_drop"
model_test.drop()
base.cursor.execute(
"SELECT model_name FROM models WHERE model_name = 'logreg_model_test_drop'"
)
assert base.cursor.fetchone() is None
def test_drop(self):
current_cursor().execute("DROP MODEL IF EXISTS logreg_model_test_drop")
model_test = LogisticRegression("logreg_model_test_drop",)
model_test.fit("public.titanic", ["age", "fare"], "survived")
current_cursor().execute(
"SELECT model_name FROM models WHERE model_name = 'logreg_model_test_drop'"
)
assert current_cursor().fetchone()[0] == "logreg_model_test_drop"
model_test.drop()
current_cursor().execute(
"SELECT model_name FROM models WHERE model_name = 'logreg_model_test_drop'"
)
assert current_cursor().fetchone() is None
def train(churn):
drop(name = "public.churn_model")
model = LogisticRegression("churn_model",
penalty = 'L2',
tol = 1e-6,
max_iter = 1000,
solver = "BFGS")
# print("Running cross_validate function\n")
# cross_validate(model, churn, churn.get_columns(exclude_columns = ["churn"]), 'churn')
print("Fitting logistic regression model...")
model.fit(churn,
churn.get_columns(exclude_columns = ["churn", "customerID"]),
'churn')
print("Success! demo.churn_model created")
print("Model AUC: " + str(model.score(method="auc")) + '\n')
# Begin Predict
model.predict(churn,
X = churn.get_columns(exclude_columns = ["churn", "customerID"]),
name = 'pred_probs')
churn.sort({"pred_probs":"desc"})
churn['pred_probs'].dropna()
return(churn)
def model(base, titanic_vd):
base.cursor.execute("DROP MODEL IF EXISTS logreg_model_test")
model_class = LogisticRegression("logreg_model_test", cursor=base.cursor)
model_class.fit("public.titanic", ["age", "fare"], "survived")
yield model_class
model_class.drop()
def model(titanic_vd):
model_class = LogisticRegression("logreg_model_test",)
model_class.drop()
model_class.fit("public.titanic", ["age", "fare"], "survived")
yield model_class
model_class.drop()