def test_sklearn_ohe_with_crossvalidation():
"""
Created 2022-02-14 to test fix to issue # 368
"""
# Set up test pipeline with wrapped OneHotEncoder, with simple regression model
# to be able to run cross-validation; use sklearn CA housing data
df = fetch_california_housing(as_frame=True).frame
y = df["MedHouseVal"]
X = (df[["HouseAge", "AveBedrms"]].assign(AveBedrms_cat=lambda x: pd.cut(
x.AveBedrms, [0, 1, 2, 3, 4, np.inf]).astype(str)).drop(
columns="AveBedrms"))
pipeline: Pipeline = Pipeline(steps=[
(
"encode_cat",
SklearnTransformerWrapper(
transformer=OneHotEncoder(drop="first", sparse=False),
variables=["AveBedrms_cat"],
),
),
("cleanup", DropFeatures(["AveBedrms_cat"])),
("model", Lasso()),
])
# Run cross-validation
results: np.ndarray = cross_val_score(pipeline,
X,
y,
scoring="neg_mean_squared_error",
cv=3)
assert not any([np.isnan(i) for i in results])
def test_drop_2_variables_integer_colnames(df_numeric_columns):
transformer = DropFeatures(features_to_drop=[0, 1])
X = transformer.fit_transform(df_numeric_columns)
# expected result
df = pd.DataFrame(
{
2: [20, 21, 19, 18],
3: [0.9, 0.8, 0.7, 0.6],
4: pd.date_range("2020-02-24", periods=4, freq="T"),
}
)
# init params
assert transformer.features_to_drop == [0, 1]
# transform params
pd.testing.assert_frame_equal(X, df)
def test_drop_2_variables(df_vartypes):
transformer = DropFeatures(features_to_drop=["City", "dob"])
X = transformer.fit_transform(df_vartypes)
# expected result
df = pd.DataFrame(
{
"Name": ["tom", "nick", "krish", "jack"],
"Age": [20, 21, 19, 18],
"Marks": [0.9, 0.8, 0.7, 0.6],
}
)
# init params
assert transformer.features_to_drop == ["City", "dob"]
# transform params
assert X.shape == (4, 3)
assert type(X) == pd.DataFrame
pd.testing.assert_frame_equal(X, df)
def test_drop_1_variable(df_vartypes):
transformer = DropFeatures(features_to_drop="City")
X = transformer.fit_transform(df_vartypes)
# expected result
df = pd.DataFrame(
{
"Name": ["tom", "nick", "krish", "jack"],
"Age": [20, 21, 19, 18],
"Marks": [0.9, 0.8, 0.7, 0.6],
"dob": pd.date_range("2020-02-24", periods=4, freq="T"),
}
)
# init params
assert transformer.features_to_drop == "City"
# transform params
assert X.shape == (4, 4)
assert type(X) == pd.DataFrame
pd.testing.assert_frame_equal(X, df)
def test_error_if_empty_list(df_vartypes):
# test case 6: passing an empty list
with pytest.raises(ValueError):
transformer = DropFeatures(features_to_drop=[])
transformer.fit_transform(df_vartypes)
def test_error_when_returning_empty_dataframe(df_vartypes):
# test case 5: dropping all columns produces warning check
with pytest.raises(ValueError):
transformer = DropFeatures(features_to_drop=list(df_vartypes.columns))
transformer.fit_transform(df_vartypes)
def test_error_if_non_existing_variables(df_vartypes):
# test case 2: passing variables that doesn't exist
with pytest.raises(KeyError):
transformer = DropFeatures(features_to_drop=["last_name"])
transformer.fit_transform(df_vartypes)
DropCorrelatedFeatures,
DropDuplicateFeatures,
DropFeatures,
DropHighPSIFeatures,
RecursiveFeatureAddition,
RecursiveFeatureElimination,
SelectByShuffling,
SelectBySingleFeaturePerformance,
SelectByTargetMeanPerformance,
SmartCorrelatedSelection,
)
_logreg = LogisticRegression(C=0.0001, max_iter=2, random_state=1)
_estimators = [
DropFeatures(features_to_drop=["0"]),
DropConstantFeatures(missing_values="ignore"),
DropDuplicateFeatures(),
DropCorrelatedFeatures(),
DropHighPSIFeatures(bins=5),
SmartCorrelatedSelection(),
SelectByShuffling(estimator=_logreg, scoring="accuracy"),
SelectByTargetMeanPerformance(bins=3, regression=False),
SelectBySingleFeaturePerformance(estimator=_logreg, scoring="accuracy"),
RecursiveFeatureAddition(estimator=_logreg, scoring="accuracy"),
RecursiveFeatureElimination(estimator=_logreg, scoring="accuracy", threshold=-100),
]
_multivariate_estimators = [
DropDuplicateFeatures(),
DropCorrelatedFeatures(),
print(mean_imputer.imputer_dict_)
X_train = mean_imputer.transform(X_train)
X_test = mean_imputer.transform(X_test)
# %% Varief whether there are missing value.
X_train[cat_vars_with_na].isnull().sum()
[var for var in cat_vars_with_na if X_test[var].isnull().sum() > 0]
#%% Temporal variables.
def elapsed_years(df, var):
df[var] = df['YrSold'] - df[var]
return df
for var in ['YearBuilt', 'YearRemodAdd', 'GarageYrBlt']：
X_train = elapsed_years(X_train, var)
X_test = elapsed_years(X_test, var)
# now we drop YrSold.
drop_features = DropFeatures(features_to_drop=['YrSold'])
X_train = mean_imputer.fit_transform(X_train)
X_test = mean_imputer.transform(X_test)
# %% Numerical variable -- transformation.
log_transformer = LogTransformer(
variables=["LotFrontage", "1stFlrSF", "GrLivArea"],
)
X_train = log_transformer.fit_transform(X_train)
X_test = log_transformer.transform(X_test)
# check that test set does not contain null values in the engineered variables
[var for var in ["LotFrontage", "1stFlrSF", "GrLivArea"] if X_test[var].isnull().sum() > 0]
# transformers
@parametrize_with_checks([
BoxCoxTransformer(),
LogTransformer(),
PowerTransformer(),
ReciprocalTransformer(),
YeoJohnsonTransformer(),
])
def test_sklearn_compatible_transformer(estimator, check):
check(estimator)
# selectors
@parametrize_with_checks([
DropFeatures(features_to_drop=["0"]),
DropConstantFeatures(missing_values="ignore"),
DropDuplicateFeatures(),
DropCorrelatedFeatures(),
SmartCorrelatedSelection(),
DropHighPSIFeatures(bins=5),
SelectByShuffling(LogisticRegression(max_iter=2, random_state=1),
scoring="accuracy"),
SelectBySingleFeaturePerformance(LogisticRegression(max_iter=2,
random_state=1),
scoring="accuracy"),
RecursiveFeatureAddition(LogisticRegression(max_iter=2, random_state=1),
scoring="accuracy"),
RecursiveFeatureElimination(
LogisticRegression(max_iter=2, random_state=1),
scoring="accuracy",
def test_error_if_fit_input_not_dataframe():
# test case 3: passing a different input than dataframe
with pytest.raises(TypeError):
transformer = DropFeatures(features_to_drop=["Name"])
transformer.fit({"Name": ["Karthik"]})
"mean_imputation",
MeanMedianImputer(
imputation_method="mean",
variables=config.model_config.numerical_vars_with_na,
),
),
# == TEMPORAL VARIABLES ====
(
"elapsed_time",
pp.TemporalVariableTransformer(
variables=config.model_config.temporal_vars,
reference_variable=config.model_config.ref_var,
),
),
("drop_features",
DropFeatures(features_to_drop=[config.model_config.ref_var])),
# ==== VARIABLE TRANSFORMATION =====
("log", LogTransformer(variables=config.model_config.numericals_log_vars)),
(
"binarizer",
SklearnTransformerWrapper(
transformer=Binarizer(threshold=0),
variables=config.model_config.binarize_vars,
),
),
# === mappers ===
(
"mapper_qual",
pp.Mapper(
variables=config.model_config.qual_vars,
mappings=config.model_config.qual_mappings,
def test_non_fitted_error(df_numeric_columns):
# test case 8: when fit is not called prior to transform
with pytest.raises(NotFittedError):
transformer = DropFeatures(features_to_drop=[0, 1])
transformer.transform(df_numeric_columns)