def test_featurize_prediction_data_in_parallel():
"""Test parallel featurization of prediction TS data"""
disco_running = util.check_disco_running()
if not disco_running:
print("Disco not running - aborting test.")
return
generate_model()
shutil.copy(pjoin(DATA_PATH, "215153_215176_218272_218934.tar.gz"),
cfg.UPLOAD_FOLDER)
shutil.copy(pjoin(DATA_PATH, "testfeature1.py"),
pjoin(cfg.CUSTOM_FEATURE_SCRIPT_FOLDER,
"TESTRUN_CF.py"))
features_and_tsdata_dict = prl_proc.featurize_prediction_data_in_parallel(
os.path.join(DATA_PATH, "215153_215176_218272_218934.tar.gz"),
"TEMP_TEST01")
print(features_and_tsdata_dict)
for fname in (
pjoin(cfg.UPLOAD_FOLDER,
"215153_215176_218272_218934.tar.gz"),
pjoin(cfg.FEATURES_FOLDER, "TESTRUN_features.csv"),
pjoin(cfg.FEATURES_FOLDER, "TESTRUN_classes.npy"),
pjoin(cfg.MODELS_FOLDER, "TESTRUN_RF.pkl"),
pjoin(cfg.CUSTOM_FEATURE_SCRIPT_FOLDER, "TESTRUN_CF.py")):
if os.path.exists(fname):
os.remove(fname)
assert "std_err" in \
features_and_tsdata_dict["dotastro_218934.dat"]["features_dict"]
def test_featurize_in_parallel():
"""Test main parallelized featurization function"""
if not util.check_disco_running():
print("Disco not running - aborting test.")
return
fname_features_dict = prl_proc.featurize_in_parallel(
pjoin(DATA_PATH,
"asas_training_subset_classes.dat"),
pjoin(DATA_PATH,
"asas_training_subset.tar.gz"),
features_to_use=["std_err", "freq1_harmonics_freq_0"],
is_test=True, custom_script_path=None)
print(fname_features_dict)
assert isinstance(fname_features_dict, dict)
for k, v in fname_features_dict.items():
assert "std_err" in v and "freq1_harmonics_freq_0" in v
