def test_matrix_filter_housing(self):
X_train, X_test, y_train, y_test, feature_names = create_housing_data()
metrics = [Metrics.MEAN_SQUARED_ERROR, Metrics.MEAN_ABSOLUTE_ERROR]
for metric in metrics:
model_task = ModelTask.REGRESSION
run_error_analyzer_on_models(X_train,
y_train,
X_test,
y_test,
feature_names,
model_task,
metric=metric)
# Test with single feature instead of two features
run_error_analyzer_on_models(X_train,
y_train,
X_test,
y_test,
feature_names,
model_task,
matrix_features=[feature_names[0]],
metric=metric)
# Note: Third feature has few unique values, tests code path
# without binning data
run_error_analyzer_on_models(X_train,
y_train,
X_test,
y_test,
feature_names,
model_task,
matrix_features=[feature_names[3]],
metric=metric)
def test_error_report_housing(self):
X_train, X_test, y_train, y_test, feature_names = \
create_housing_data()
models = create_models_regression(X_train, y_train)
for model in models:
categorical_features = []
run_error_analyzer(model, X_test, y_test, feature_names,
categorical_features)
def test_error_report_housing_pandas(self, filter_features):
X_train, X_test, y_train, y_test, feature_names = \
create_housing_data()
X_train = create_dataframe(X_train, feature_names)
X_test = create_dataframe(X_test, feature_names)
models = create_models_regression(X_train, y_train)
for model in models:
categorical_features = []
run_error_analyzer(model,
X_test,
y_test,
feature_names,
categorical_features,
filter_features=filter_features)
def test_matrix_filter_housing_quantile_binning(self):
# Test quantile binning on CRIM feature in california housing dataset,
# which errored out due to first category not fitting into bins
(X_train, X_test, y_train, y_test,
feature_names) = create_housing_data(test_size=0.5)
model_task = ModelTask.REGRESSION
matrix_features = ['Population']
run_error_analyzer_on_models(X_train,
y_train,
X_test,
y_test,
feature_names,
model_task,
matrix_features=matrix_features,
quantile_binning=True)
def test_matrix_filter_housing_filters(self):
X_train, X_test, y_train, y_test, feature_names = create_housing_data()
filters = [{
'arg': [600],
'column': 'Population',
'method': 'less and equal'
}, {
'arg': [6],
'column': 'AveRooms',
'method': 'greater'
}]
model_task = ModelTask.REGRESSION
run_error_analyzer_on_models(X_train,
y_train,
X_test,
y_test,
feature_names,
model_task,
filters=filters)
