def test_get_hourly_chunks_6(self):
"""Ensures correct output from get_hourly_chunks.
In this case there are 6 hours per chunk.
"""
this_chunk_to_hours_dict = temporal_subsetting.get_hourly_chunks(
num_hours_per_chunk=6, verbose=False)
self.assertTrue(
_compare_chunk_dicts(this_chunk_to_hours_dict,
CHUNK_TO_HOURS_DICT_6EACH))
def _plot_by_hour(evaluation_dir_name, num_hours_per_chunk, confidence_level,
output_dir_name):
"""Plots model evaluation by hour.
:param evaluation_dir_name: See documentation at top of file.
:param num_hours_per_chunk: Same.
:param confidence_level: Same.
:param output_dir_name: Same.
:return: output_file_names: Paths to figures saved by this method.
"""
chunk_to_hours_dict = temporal_subsetting.get_hourly_chunks(
num_hours_per_chunk=num_hours_per_chunk, verbose=False)
num_bootstrap_reps = None
num_chunks = len(chunk_to_hours_dict.keys())
auc_matrix = numpy.full((num_chunks, 3), numpy.nan)
pod_matrix = numpy.full((num_chunks, 3), numpy.nan)
far_matrix = numpy.full((num_chunks, 3), numpy.nan)
csi_matrix = numpy.full((num_chunks, 3), numpy.nan)
num_examples_by_chunk = numpy.full(num_chunks, 0, dtype=int)
num_positive_ex_by_chunk = numpy.full(num_chunks, 0, dtype=int)
for i in range(num_chunks):
this_eval_file_name = model_eval.find_file(
directory_name=evaluation_dir_name,
hours_in_subset=chunk_to_hours_dict[i],
raise_error_if_missing=False)
if not os.path.isfile(this_eval_file_name):
warning_string = (
'Cannot find file (this may or may not be a problem). Expected'
' at: "{0:s}"').format(this_eval_file_name)
warnings.warn(warning_string)
continue
print('Reading data from: "{0:s}"...'.format(this_eval_file_name))
this_evaluation_dict = model_eval.read_evaluation(this_eval_file_name)
num_examples_by_chunk[i] = len(
this_evaluation_dict[model_eval.OBSERVED_LABELS_KEY])
num_positive_ex_by_chunk[i] = numpy.sum(
this_evaluation_dict[model_eval.OBSERVED_LABELS_KEY])
this_evaluation_table = this_evaluation_dict[
model_eval.EVALUATION_TABLE_KEY]
this_num_bootstrap_reps = len(this_evaluation_table.index)
if num_bootstrap_reps is None:
num_bootstrap_reps = this_num_bootstrap_reps
assert num_bootstrap_reps == this_num_bootstrap_reps
these_auc = this_evaluation_table[model_eval.AUC_KEY].values
these_pod = this_evaluation_table[model_eval.POD_KEY].values
these_far = (
1. - this_evaluation_table[model_eval.SUCCESS_RATIO_KEY].values)
these_csi = this_evaluation_table[model_eval.CSI_KEY].values
auc_matrix[i, 1] = numpy.nanmean(these_auc)
pod_matrix[i, 1] = numpy.nanmean(these_pod)
far_matrix[i, 1] = numpy.nanmean(these_far)
csi_matrix[i, 1] = numpy.nanmean(these_csi)
auc_matrix[i,
0], auc_matrix[i,
2] = (bootstrapping.get_confidence_interval(
stat_values=these_auc,
confidence_level=confidence_level))
pod_matrix[i,
0], pod_matrix[i,
2] = (bootstrapping.get_confidence_interval(
stat_values=these_pod,
confidence_level=confidence_level))
far_matrix[i,
0], far_matrix[i,
2] = (bootstrapping.get_confidence_interval(
stat_values=these_far,
confidence_level=confidence_level))
csi_matrix[i,
0], csi_matrix[i,
2] = (bootstrapping.get_confidence_interval(
stat_values=these_csi,
confidence_level=confidence_level))
x_tick_labels = [None] * num_chunks
x_tick_values = numpy.linspace(0,
num_chunks - 1,
num=num_chunks,
dtype=float)
for i in range(num_chunks):
these_hours = chunk_to_hours_dict[i]
if len(these_hours) == 1:
x_tick_labels[i] = '{0:02d}'.format(these_hours[0])
else:
x_tick_labels[i] = '{0:02d}-{1:02d}'.format(
numpy.min(these_hours), numpy.max(these_hours))
figure_object, axes_object = _plot_auc_and_csi(
auc_matrix=auc_matrix,
csi_matrix=csi_matrix,
num_examples_by_chunk=num_examples_by_chunk,
num_bootstrap_reps=num_bootstrap_reps,
plot_legend=False)
axes_object.set_xticks(x_tick_values)
axes_object.set_xticklabels(x_tick_labels, rotation=90.)
axes_object.set_xlabel('Hour (UTC)')
plotting_utils.label_axes(axes_object=axes_object,
label_string='(b)',
x_coord_normalized=-0.075,
y_coord_normalized=1.02)
auc_csi_file_name = '{0:s}/hourly_auc_and_csi.jpg'.format(output_dir_name)
print('Saving figure to: "{0:s}"...'.format(auc_csi_file_name))
figure_object.savefig(auc_csi_file_name,
dpi=FIGURE_RESOLUTION_DPI,
pad_inches=0,
bbox_inches='tight')
pyplot.close(figure_object)
figure_object, axes_object = _plot_pod_and_far(
pod_matrix=pod_matrix,
far_matrix=far_matrix,
num_positive_ex_by_chunk=num_positive_ex_by_chunk,
num_bootstrap_reps=num_bootstrap_reps,
plot_legend=False)
axes_object.set_xticks(x_tick_values)
axes_object.set_xticklabels(x_tick_labels, rotation=90.)
axes_object.set_xlabel('Hour (UTC)')
plotting_utils.label_axes(axes_object=axes_object,
label_string='(d)',
x_coord_normalized=-0.075,
y_coord_normalized=1.02)
pod_far_file_name = '{0:s}/hourly_pod_and_far.jpg'.format(output_dir_name)
print('Saving figure to: "{0:s}"...'.format(pod_far_file_name))
figure_object.savefig(pod_far_file_name,
dpi=FIGURE_RESOLUTION_DPI,
pad_inches=0,
bbox_inches='tight')
pyplot.close(figure_object)
return [auc_csi_file_name, pod_far_file_name]
def _run(input_file_name, num_months_per_chunk, num_hours_per_chunk,
output_dir_name):
"""Subsets ungridded predictions by time.
This is effectively the main method.
:param input_file_name: See documentation at top of file.
:param num_months_per_chunk: Same.
:param num_hours_per_chunk: Same.
:param output_dir_name: Same.
"""
if num_months_per_chunk > 0:
chunk_to_months_dict = temporal_subsetting.get_monthly_chunks(
num_months_per_chunk=num_months_per_chunk, verbose=True)
num_monthly_chunks = len(chunk_to_months_dict.keys())
print(SEPARATOR_STRING)
else:
num_monthly_chunks = 0
if num_hours_per_chunk > 0:
chunk_to_hours_dict = temporal_subsetting.get_hourly_chunks(
num_hours_per_chunk=num_hours_per_chunk, verbose=True)
num_hourly_chunks = len(chunk_to_hours_dict.keys())
print(SEPARATOR_STRING)
else:
num_hourly_chunks = 0
print('Reading input data from: "{0:s}"...'.format(input_file_name))
prediction_dict = prediction_io.read_ungridded_predictions(input_file_name)
storm_times_unix_sec = prediction_dict[prediction_io.STORM_TIMES_KEY]
storm_months = None
for i in range(num_monthly_chunks):
these_storm_indices, storm_months = (
temporal_subsetting.get_events_in_months(
event_months=storm_months,
event_times_unix_sec=storm_times_unix_sec,
desired_months=chunk_to_months_dict[i],
verbose=True))
this_prediction_dict = prediction_io.subset_ungridded_predictions(
prediction_dict=prediction_dict,
desired_storm_indices=these_storm_indices)
this_output_file_name = prediction_io.find_ungridded_file(
directory_name=output_dir_name,
months_in_subset=chunk_to_months_dict[i],
raise_error_if_missing=False)
print('Writing temporal subset to: "{0:s}"...'.format(
this_output_file_name))
prediction_io.write_ungridded_predictions(
netcdf_file_name=this_output_file_name,
class_probability_matrix=this_prediction_dict[
prediction_io.PROBABILITY_MATRIX_KEY],
storm_ids=this_prediction_dict[prediction_io.STORM_IDS_KEY],
storm_times_unix_sec=this_prediction_dict[
prediction_io.STORM_TIMES_KEY],
observed_labels=this_prediction_dict[
prediction_io.OBSERVED_LABELS_KEY],
target_name=this_prediction_dict[prediction_io.TARGET_NAME_KEY],
model_file_name=this_prediction_dict[prediction_io.MODEL_FILE_KEY])
print(SEPARATOR_STRING)
storm_hours = None
for i in range(num_hourly_chunks):
these_storm_indices, storm_hours = (
temporal_subsetting.get_events_in_hours(
event_hours=storm_hours,
event_times_unix_sec=storm_times_unix_sec,
desired_hours=chunk_to_hours_dict[i],
verbose=True))
if len(these_storm_indices) == 0:
continue
this_prediction_dict = prediction_io.subset_ungridded_predictions(
prediction_dict=prediction_dict,
desired_storm_indices=these_storm_indices)
this_output_file_name = prediction_io.find_ungridded_file(
directory_name=output_dir_name,
hours_in_subset=chunk_to_hours_dict[i],
raise_error_if_missing=False)
print('Writing temporal subset to: "{0:s}"...'.format(
this_output_file_name))
prediction_io.write_ungridded_predictions(
netcdf_file_name=this_output_file_name,
class_probability_matrix=this_prediction_dict[
prediction_io.PROBABILITY_MATRIX_KEY],
storm_ids=this_prediction_dict[prediction_io.STORM_IDS_KEY],
storm_times_unix_sec=this_prediction_dict[
prediction_io.STORM_TIMES_KEY],
observed_labels=this_prediction_dict[
prediction_io.OBSERVED_LABELS_KEY],
target_name=this_prediction_dict[prediction_io.TARGET_NAME_KEY],
model_file_name=this_prediction_dict[prediction_io.MODEL_FILE_KEY])
if i != num_hourly_chunks - 1:
print(SEPARATOR_STRING)