def test_save_outliers(test_config: PlotCrossValidationConfig,
test_output_dirs: OutputFolderForTests) -> None:
"""Test to make sure the outlier file for a split is as expected"""
test_config.outputs_directory = test_output_dirs.root_dir
test_config.outlier_range = 0
assert test_config.run_recovery_id
dataset_split_metrics = {
x: _get_metrics_df(test_config.run_recovery_id, x)
for x in [ModelExecutionMode.VAL]
}
save_outliers(test_config, dataset_split_metrics,
test_config.outputs_directory)
filename = f"{ModelExecutionMode.VAL.value}_outliers.txt"
assert_text_files_match(full_file=test_config.outputs_directory / filename,
expected_file=full_ml_test_data_path(filename))
# Now test without the CSV_INSTITUTION_HEADER and CSV_SERIES_HEADER columns, which will be missing in institutions' environments
dataset_split_metrics_pruned = {
x: _get_metrics_df(test_config.run_recovery_id, x).drop(
columns=[CSV_INSTITUTION_HEADER, CSV_SERIES_HEADER],
errors="ignore")
for x in [ModelExecutionMode.VAL]
}
save_outliers(test_config, dataset_split_metrics_pruned,
test_config.outputs_directory)
test_data_filename = f"{ModelExecutionMode.VAL.value}_outliers_pruned.txt"
assert_text_files_match(
full_file=test_config.outputs_directory / filename,
expected_file=full_ml_test_data_path(test_data_filename))
def test_save_outliers(test_config_ensemble: PlotCrossValidationConfig,
test_output_dirs: OutputFolderForTests) -> None:
"""Test to make sure the outlier file for a split is as expected"""
test_config_ensemble.outputs_directory = test_output_dirs.root_dir
test_config_ensemble.outlier_range = 0
dataset_split_metrics = {x: _get_metrics_df(x) for x in [ModelExecutionMode.VAL]}
save_outliers(test_config_ensemble, dataset_split_metrics, test_config_ensemble.outputs_directory)
f = f"{ModelExecutionMode.VAL.value}_outliers.txt"
assert_text_files_match(full_file=test_config_ensemble.outputs_directory / f,
expected_file=full_ml_test_data_path(f))
def test_save_outliers(test_config_ensemble: PlotCrossValidationConfig,
test_output_dirs: TestOutputDirectories) -> None:
"""Test to make sure the outlier file for a split is as expected"""
test_config_ensemble.outputs_directory = test_output_dirs.root_dir
test_config_ensemble.outlier_range = 0
dataset_split_metrics = {x: _get_metrics_df(x) for x in [ModelExecutionMode.VAL]}
save_outliers(test_config_ensemble, dataset_split_metrics, Path(test_config_ensemble.outputs_directory))
assert_file_contents_match_exactly(full_file=Path(test_config_ensemble.outputs_directory)
/ f"{ModelExecutionMode.VAL.value}_outliers.txt",
expected_file=Path(
full_ml_test_data_path(
f"{ModelExecutionMode.VAL.value}_outliers.txt")))
