def test_cohort_filter_classification_outcome(self):
X_train, X_test, y_train, y_test, numeric, categorical = \
create_simple_titanic_data()
feature_names = categorical + numeric
clf = create_titanic_pipeline(X_train, y_train)
categorical_features = categorical
# the indexes 1, 2 correspond to false positives and false negatives
filters = [{
'arg': [1, 2],
'column': CLASSIFICATION_OUTCOME,
'method': 'includes'
}]
pred_y = clf.predict(X_test)
validation_data = create_validation_data(X_test, y_test, pred_y)
validation_filter = validation_data[PRED_Y] != validation_data[TRUE_Y]
validation_data = validation_data.loc[validation_filter]
validation_data = validation_data.drop(columns=PRED_Y)
model_task = ModelTask.CLASSIFICATION
run_error_analyzer(validation_data,
clf,
X_test,
y_test,
feature_names,
categorical_features,
model_task,
filters=filters)
def test_importances_titanic(self):
X_train, X_test, y_train, y_test, numeric, categorical = \
create_simple_titanic_data()
feature_names = categorical + numeric
clf = create_titanic_pipeline(X_train, y_train)
categorical_features = categorical
run_error_analyzer(clf, X_test, y_test, feature_names,
categorical_features)
def test_matrix_filter_titanic(self):
x_train, x_test, y_train, y_test, numeric, categorical = \
create_simple_titanic_data()
feature_names = categorical + numeric
clf = create_titanic_pipeline(x_train, y_train)
categorical_features = categorical
run_error_analyzer(clf, x_test, y_test, feature_names,
categorical_features)
def titanic_simple():
x_train, x_test, y_train, y_test, numeric, categorical = create_simple_titanic_data()
yield {
DatasetConstants.X_TRAIN: x_train,
DatasetConstants.X_TEST: x_test,
DatasetConstants.Y_TRAIN: y_train,
DatasetConstants.Y_TEST: y_test,
DatasetConstants.NUMERIC: numeric,
DatasetConstants.CATEGORICAL: categorical
}
def test_matrix_filter_titanic(self):
(X_train, X_test, y_train, y_test, numeric,
categorical) = create_simple_titanic_data()
feature_names = categorical + numeric
clf = create_titanic_pipeline(X_train, y_train)
categorical_features = categorical
run_error_analyzer(clf,
X_test,
y_test,
feature_names,
categorical_features,
model_task=ModelTask.CLASSIFICATION)
def test_cohort_filter_excludes(self):
X_train, X_test, y_train, y_test, numeric, categorical = \
create_simple_titanic_data()
feature_names = categorical + numeric
clf = create_titanic_pipeline(X_train, y_train)
categorical_features = categorical
# the indexes other than 0, 2 correspond to Q
filters = [{'arg': [0, 2], 'column': EMBARKED, 'method': 'excludes'}]
validation_data = create_validation_data(X_test, y_test)
filter_embarked = X_test[EMBARKED].isin(['Q'])
validation_data = validation_data.loc[filter_embarked]
model_task = ModelTask.CLASSIFICATION
run_error_analyzer(validation_data,
clf,
X_test,
y_test,
feature_names,
categorical_features,
model_task,
filters=filters)
def test_invalid_comparison_titanic(self, analyzer_type):
(X_train, X_test, y_train, y_test, numeric,
categorical) = create_simple_titanic_data()
tree_features = [STRING_INDEX]
feature_names = categorical + numeric + tree_features
# Create a bad dummy string categorical feature
X_train = add_string_index_col(X_train)
X_test = add_string_index_col(X_test)
clf = create_titanic_pipeline(X_train, y_train)
categorical_features = categorical
tree_features = tree_features + numeric
with pytest.raises(TypeError) as ve:
run_error_analyzer(clf,
X_test,
y_test,
feature_names,
analyzer_type,
categorical_features,
tree_features,
model_task=ModelTask.CLASSIFICATION)
assert ('Column string_index of type string is incorrectly treated '
'as numeric with threshold value') in str(ve.value)
