def model(winequality_vd):
current_cursor().execute("DROP MODEL IF EXISTS norm_model_test")
model_class = Normalizer("norm_model_test", )
model_class.fit("public.winequality",
["citric_acid", "residual_sugar", "alcohol"])
yield model_class
model_class.drop()
def test_model_from_vDF(self, winequality_vd):
current_cursor().execute("DROP MODEL IF EXISTS norm_vDF")
model_test = Normalizer("norm_vDF", )
model_test.fit(winequality_vd, ["alcohol", "quality"])
current_cursor().execute(
"SELECT model_name FROM models WHERE model_name = 'norm_vDF'")
assert current_cursor().fetchone()[0] == "norm_vDF"
model_test.drop()
def test_model_from_vDF(self, base, winequality_vd):
base.cursor.execute("DROP MODEL IF EXISTS norm_vDF")
model_test = Normalizer("norm_vDF", cursor=base.cursor)
model_test.fit(winequality_vd, ["alcohol", "quality"])
base.cursor.execute(
"SELECT model_name FROM models WHERE model_name = 'norm_vDF'")
assert base.cursor.fetchone()[0] == "norm_vDF"
model_test.drop()
def test_set_cursor(self, base):
model_test = Normalizer("norm_cursor_test", cursor=base.cursor)
# TODO: creat a new cursor
model_test.set_cursor(base.cursor)
model_test.drop()
model_test.fit("public.winequality", ["alcohol"])
base.cursor.execute(
"SELECT model_name FROM models WHERE model_name = 'norm_cursor_test'"
)
assert base.cursor.fetchone()[0] == "norm_cursor_test"
model_test.drop()
def test_to_python(self, model):
# Zscore
current_cursor().execute(
"SELECT APPLY_NORMALIZE(citric_acid, residual_sugar, alcohol USING PARAMETERS model_name = '{}', match_by_pos=True) FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(model.name))
prediction = current_cursor().fetchone()[0]
assert prediction == pytest.approx(
model.to_python(return_str=False)([[3.0, 11.0, 93.0]])[0][0])
# Minmax
model2 = Normalizer("norm_model_test2", method="minmax")
model2.drop()
model2.fit("public.winequality",
["citric_acid", "residual_sugar", "alcohol"])
current_cursor().execute(
"SELECT APPLY_NORMALIZE(citric_acid, residual_sugar, alcohol USING PARAMETERS model_name = '{}', match_by_pos=True) FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(model2.name))
prediction = current_cursor().fetchone()[0]
assert prediction == pytest.approx(
model2.to_python(return_str=False)([[3.0, 11.0, 93.0]])[0][0])
model2.drop()
# Robust Zscore
model3 = Normalizer("norm_model_test2", method="robust_zscore")
model3.drop()
model3.fit("public.winequality",
["citric_acid", "residual_sugar", "alcohol"])
current_cursor().execute(
"SELECT APPLY_NORMALIZE(citric_acid, residual_sugar, alcohol USING PARAMETERS model_name = '{}', match_by_pos=True) FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(model3.name))
prediction = current_cursor().fetchone()[0]
assert prediction == pytest.approx(
model3.to_python(return_str=False)([[3.0, 11.0, 93.0]])[0][0])
model3.drop()
def test_drop(self):
current_cursor().execute("DROP MODEL IF EXISTS norm_model_test_drop")
model_test = Normalizer("norm_model_test_drop", )
model_test.fit("public.winequality", ["alcohol", "quality"])
current_cursor().execute(
"SELECT model_name FROM models WHERE model_name = 'norm_model_test_drop'"
)
assert current_cursor().fetchone()[0] == "norm_model_test_drop"
model_test.drop()
current_cursor().execute(
"SELECT model_name FROM models WHERE model_name = 'norm_model_test_drop'"
)
assert current_cursor().fetchone() is None
def test_drop(self, base):
base.cursor.execute("DROP MODEL IF EXISTS norm_model_test_drop")
model_test = Normalizer("norm_model_test_drop", cursor=base.cursor)
model_test.fit("public.winequality", ["alcohol", "quality"])
base.cursor.execute(
"SELECT model_name FROM models WHERE model_name = 'norm_model_test_drop'"
)
assert base.cursor.fetchone()[0] == "norm_model_test_drop"
model_test.drop()
base.cursor.execute(
"SELECT model_name FROM models WHERE model_name = 'norm_model_test_drop'"
)
assert base.cursor.fetchone() is None
def test_to_memmodel(self, model):
# Zscore
current_cursor().execute(
"SELECT APPLY_NORMALIZE(citric_acid, residual_sugar, alcohol USING PARAMETERS model_name = '{}', match_by_pos=True) FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(model.name))
prediction = [float(elem) for elem in current_cursor().fetchone()]
current_cursor().execute(
"SELECT {} FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(", ".join(model.to_memmodel().transform_sql(
["citric_acid", "residual_sugar", "alcohol"]))))
prediction2 = [float(elem) for elem in current_cursor().fetchone()]
assert prediction == pytest.approx(prediction2)
prediction3 = model.to_memmodel().transform([[3.0, 11.0, 93.0]])
assert prediction[0] == pytest.approx(prediction3[0][0])
assert prediction[1] == pytest.approx(prediction3[0][1])
assert prediction[2] == pytest.approx(prediction3[0][2])
# Minmax
model2 = Normalizer("norm_model_test2", method="minmax")
model2.drop()
model2.fit("public.winequality",
["citric_acid", "residual_sugar", "alcohol"])
current_cursor().execute(
"SELECT APPLY_NORMALIZE(citric_acid, residual_sugar, alcohol USING PARAMETERS model_name = '{}', match_by_pos=True) FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(model2.name))
prediction = [float(elem) for elem in current_cursor().fetchone()]
current_cursor().execute(
"SELECT {} FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(", ".join(model2.to_memmodel().transform_sql(
["citric_acid", "residual_sugar", "alcohol"]))))
prediction2 = [float(elem) for elem in current_cursor().fetchone()]
assert prediction == pytest.approx(prediction2)
prediction3 = model2.to_memmodel().transform([[3.0, 11.0, 93.0]])
assert prediction[0] == pytest.approx(prediction3[0][0])
assert prediction[1] == pytest.approx(prediction3[0][1])
assert prediction[2] == pytest.approx(prediction3[0][2])
model2.drop()
# Robust Zscore
model3 = Normalizer("norm_model_test2", method="robust_zscore")
model3.drop()
model3.fit("public.winequality",
["citric_acid", "residual_sugar", "alcohol"])
current_cursor().execute(
"SELECT APPLY_NORMALIZE(citric_acid, residual_sugar, alcohol USING PARAMETERS model_name = '{}', match_by_pos=True) FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(model3.name))
prediction = [float(elem) for elem in current_cursor().fetchone()]
current_cursor().execute(
"SELECT {} FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(", ".join(model3.to_memmodel().transform_sql(
["citric_acid", "residual_sugar", "alcohol"]))))
prediction2 = [float(elem) for elem in current_cursor().fetchone()]
assert prediction == pytest.approx(prediction2)
prediction3 = model3.to_memmodel().transform([[3.0, 11.0, 93.0]])
assert prediction[0] == pytest.approx(prediction3[0][0])
assert prediction[1] == pytest.approx(prediction3[0][1])
assert prediction[2] == pytest.approx(prediction3[0][2])
model3.drop()
def test_to_sql(self, model):
# Zscore
model.cursor.execute(
"SELECT APPLY_NORMALIZE(citric_acid, residual_sugar, alcohol USING PARAMETERS model_name = '{}', match_by_pos=True) FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(model.name))
prediction = [float(elem) for elem in model.cursor.fetchone()]
model.cursor.execute(
"SELECT {} FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(model.to_sql()))
prediction2 = [float(elem) for elem in model.cursor.fetchone()]
assert prediction == pytest.approx(prediction2)
# Minmax
model2 = Normalizer("norm_model_test2",
cursor=model.cursor,
method="minmax")
model2.drop()
model2.fit("public.winequality",
["citric_acid", "residual_sugar", "alcohol"])
model2.cursor.execute(
"SELECT APPLY_NORMALIZE(citric_acid, residual_sugar, alcohol USING PARAMETERS model_name = '{}', match_by_pos=True) FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(model2.name))
prediction = [float(elem) for elem in model2.cursor.fetchone()]
model2.cursor.execute(
"SELECT {} FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(model2.to_sql()))
prediction2 = [float(elem) for elem in model2.cursor.fetchone()]
assert prediction == pytest.approx(prediction2)
model2.drop()
# Robust Zscore
model3 = Normalizer("norm_model_test2",
cursor=model.cursor,
method="robust_zscore")
model3.drop()
model3.fit("public.winequality",
["citric_acid", "residual_sugar", "alcohol"])
model3.cursor.execute(
"SELECT APPLY_NORMALIZE(citric_acid, residual_sugar, alcohol USING PARAMETERS model_name = '{}', match_by_pos=True) FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(model3.name))
prediction = [float(elem) for elem in model3.cursor.fetchone()]
model3.cursor.execute(
"SELECT {} FROM (SELECT 3.0 AS citric_acid, 11.0 AS residual_sugar, 93. AS alcohol) x"
.format(model3.to_sql()))
prediction2 = [float(elem) for elem in model3.cursor.fetchone()]
assert prediction == pytest.approx(prediction2)
model3.drop()
def test_get_transform(self, winequality_vd, model):
# Zscore
winequality_trans = model.transform(
winequality_vd, X=["citric_acid", "residual_sugar", "alcohol"])
assert winequality_trans["citric_acid"].mean() == pytest.approx(
0.0, abs=1e-6)
assert winequality_trans["residual_sugar"].mean() == pytest.approx(
0.0, abs=1e-6)
assert winequality_trans["alcohol"].mean() == pytest.approx(0.0,
abs=1e-6)
# Minmax
model2 = Normalizer("norm_model_test2",
cursor=model.cursor,
method="minmax")
model2.drop()
model2.fit("public.winequality",
["citric_acid", "residual_sugar", "alcohol"])
winequality_trans = model2.transform(
winequality_vd, X=["citric_acid", "residual_sugar", "alcohol"])
assert winequality_trans["citric_acid"].min() == pytest.approx(
0.0, abs=1e-6)
assert winequality_trans["residual_sugar"].max() == pytest.approx(
1.0, abs=1e-6)
assert winequality_trans["alcohol"].min() == pytest.approx(0.0,
abs=1e-6)
model2.drop()
# Robust Zscore
model3 = Normalizer("norm_model_test2",
cursor=model.cursor,
method="robust_zscore")
model3.drop()
model3.fit("public.winequality",
["citric_acid", "residual_sugar", "alcohol"])
winequality_trans = model3.transform(
winequality_vd, X=["citric_acid", "residual_sugar", "alcohol"])
assert winequality_trans["citric_acid"].median() == pytest.approx(
0.0, abs=1e-6)
assert winequality_trans["residual_sugar"].median() == pytest.approx(
0.0, abs=1e-6)
assert winequality_trans["alcohol"].median() == pytest.approx(0.0,
abs=1e-6)
model3.drop()
def test_repr(self, model):
assert "column_name | avg |std_dev" in model.__repr__()
model_repr = Normalizer("model_repr")
model_repr.drop()
assert model_repr.__repr__() == "<Normalizer>"