y_lim=[1362, 1369],
data_points_triplets=[
(base_sig.t_a_nn, final_res.a_nn_corrected, f"{Const.A}_corrected"),
(base_sig.t_b_nn, final_res.b_nn_corrected, f"{Const.B}_corrected")
],
x_label=Const.YEAR_UNIT, y_label=Const.TSI_UNIT, max_points=1e7)
if __name__ == "__main__":
ignore_warnings()
visualizer = Visualizer()
visualizer.set_figsize()
ARGS = parse_arguments()
results_dir_path = create_results_dir(Const.RESULTS_DIR, f"{ARGS.model_type}_{ARGS.output_method}")
create_logger(results_dir_path)
mode = Mode.VIRGO
data, t_field_name, a_field_name, b_field_name, _ = load_data_run(ARGS, mode)
model, model_type, correction_method, exposure_method, output_model, output_method, outlier_fraction \
= setup_run(ARGS, mode, results_dir_path)
fitter = ModelFitter(data=data,
t_field_name=t_field_name,
a_field_name=a_field_name,
b_field_name=b_field_name,
exposure_method=exposure_method,
outlier_fraction=outlier_fraction)
def analysis_job(dataset: Dataset, model_type: str, output_model_type: str,
model_params: str, correction_method: str):
try:
# Enforce optional params
# TODO: improve this code
# TODO: this code sets constants that are used within model and passes constants that are passed via init
# TODO: this way all relevant constants are taken into account
num_inducing_points = Gpc.NUM_INDUCING_POINTS
points_in_window = Gpc.POINTS_IN_WINDOW
model_params_dict = eval(model_params)
for key in model_params_dict:
if key == "NUM_INDUCING_POINTS":
num_inducing_points = model_params_dict[key]
elif key == "POINTS_IN_WINDOW":
points_in_window = model_params_dict[key]
else:
setattr(Gpc, key, model_params_dict[key])
# Load pickle
status.update_progress("Loading dataset", 10)
pickle_location = dataset.pickle_location
with open(pickle_location, "rb") as f:
fitter: ModelFitter = pickle.load(f)
# Get models
model, output_model = get_models(model_type, output_model_type,
num_inducing_points, points_in_window)
# Run Fitter
result = fitter(model, output_model,
CorrectionMethod(correction_method), Mode.VIRGO)
# Create result folder
with app.app_context():
results_dir_path = create_results_dir(
os.path.join(current_app.root_path, "static", "results"),
f"{model_type}_{output_model_type}")
# Store signals
save_modeling_result(results_dir_path, result, model_type)
result.out.params_out.svgp_inducing_points = None
# Plot results
status.update_progress("Plotting results", 90)
visualizer = Visualizer()
visualizer.set_figsize()
plot_results(visualizer, result, results_dir_path,
f"{model_type}_{output_model_type}")
# Store folder location to status
status.set_folder(basename(results_dir_path))
# Close all figures from matplotlib
plt.close("all")
# Finish job
status.set_percentage(100)
except Exception as e:
print(e)
finally:
status.release()
if __name__ == "__main__":
ignore_warnings()
visualizer = Visualizer()
visualizer.set_figsize()
run_type = "comparison"
ARGS = parse_arguments()
if ARGS.mode == "gen":
mode = Mode.GENERATOR
run_type = "gen_" + run_type
else:
mode = Mode.VIRGO
results_dir_path = create_results_dir(Const.RESULTS_DIR, run_type)
create_logger(results_dir_path)
data, t_field_name, a_field_name, b_field_name, ground_truth = load_data_run(
ARGS, mode)
results = []
for model_type in [
"exp", "exp_lin", "spline", "isotonic", "smooth_monotonic",
"ensemble"
]:
ARGS.model_type = model_type
model, _, correction_method, exposure_method, output_model, output_method, outlier_fraction \
= setup_run(ARGS, mode, results_dir_path)
fitter = ModelFitter(data=data,
visualizer.plot_signals([(frq[frq <= frq_range], a_s[frq <= frq_range], "{}_spectrum".format(Const.A), False)],
results_dir_path, "{}_spectrum".format(Const.A), legend="upper right",
y_label=Const.SPECTRUM_UNIT)
b_s, fs, frq = fft_spectrum(b_inter, sampling_period)
frq_range = fs / 2
visualizer.plot_signals([(frq[frq <= frq_range], b_s[frq <= frq_range], "{}_spectrum".format(Const.B), False)],
results_dir_path, "{}_spectrum".format(Const.B), legend="upper right",
y_label=Const.SPECTRUM_UNIT)
if __name__ == "__main__":
ARGS = parse_arguments()
results_dir_path = create_results_dir(Const.RESULTS_DIR, "data_analysis")
create_logger(results_dir_path)
visualizer = Visualizer()
visualizer.set_figsize()
# Load data
data = None
if ARGS.data_file == "premos":
pass
elif ARGS.data_file == "virgo_tsi":
data = load_data(Const.DATA_DIR, Const.VIRGO_TSI_FILE, "virgo_tsi")
plot_base_virgo_tsi_signals(data)
elif ARGS.data_file == "spm":
data = load_data(Const.DATA_DIR, Const.SPM_FILE, "spm")
plot_base_spm_signals(data)