def load_or_create_datasets(window_ms=100,
stride_ms=100,
imp_split=True,
dos_type='original',
force_create=False,
verbose=False,
in_parallel=True):
"""Returns the training and validation sets associated with input argument combination.
If the datasets do not exist, they are created and saved in the process.
:param window_ms: the window size (int ms).
:param stride_ms: the step size (int ms).
:param imp_split: a flag indicating whether the impersonation dataset has split labels.
:param dos_type: a string indicating the DoS dataset used ('modified', 'original').
:param force_create: a flag indicating whether the dataset should be generated, even if it exists.
:param verbose: a flag indicating how much progress should be output to console.
:param in_parallel: a flag indicating whether features should be calculated in parallel.
:return: a list of training DataPoints, a list of validation DataPoints, a dictionary of feature durations.
"""
training_name, _ = __get_dataset_path(window_ms, stride_ms, imp_split,
dos_type, 'training')
validation_name, _ = __get_dataset_path(window_ms, stride_ms, imp_split,
dos_type, 'validation')
time_path, directory = __get_dataset_path(window_ms, stride_ms, imp_split,
dos_type, 'validation_time')
# Load the datasets if they exist.
if os.path.exists(training_name) and os.path.exists(
validation_name) and not force_create:
training_set = datareader_csv.load_datapoints(training_name,
verbose=verbose)
validation_set = datareader_csv.load_datapoints(validation_name,
verbose=verbose)
feature_durations = datareader_csv.load_feature_durations(time_path)
else:
# Create and save the datasets otherwise.
training_set, validation_set, feature_durations = get_mixed_training_validation(
window_ms,
stride_ms,
imp_split,
dos_type,
verbose=verbose,
in_parallel=in_parallel)
__write_datapoints_csv(training_set, window_ms, stride_ms, imp_split,
dos_type, 'training')
__write_datapoints_csv(validation_set, window_ms, stride_ms, imp_split,
dos_type, 'validation')
datawriter_csv.save_feature_durations(feature_durations, time_path,
directory)
return training_set, validation_set, feature_durations
def get_result_feature_breakdown(result: Result, type='validation'):
"""
Gets a dict mapping feature labels to feature times. This is loaded according to a specific Result
:param result: The result to load feature times to
:param type: The type of feature times to load. Valid values are 'validation' and 'test'
:return:
"""
imp_name = "imp_split" if result.imp_split else "imp_full"
name = f"mixed_{type}_time_{result.window_ms}ms_{result.stride_ms}ms"
directory = f"data/feature/{imp_name}/{result.dos_type}/"
path = directory + name + ".csv"
return datareader_csv.load_feature_durations(path)
result.imp_split, result.dos_type, result.model,
result.baseline, result.subset, True)
for result in best_validation_results
]
plot_types = ['f1', 'fpr', 'fnr', 'recall', 'precision', 'accuracy']
metrics_types = ['macro', 'normal', 'impersonation', 'dos', 'fuzzy']
for plot_type in plot_types:
for metrics_type in metrics_types:
plot_barchart_results(best_test_results, plot_type, metrics_type)
plot_barchart_results(best_test_results, 'model_time')
plot_barchart_feature_results(best_test_results)
durations_path = f"data\\feature\\{conf.imp_type}\\{conf.dos_type}\\mixed_validation_time_100ms_100ms.csv"
feature_times = datareader_csv.load_feature_durations(durations_path)
del feature_times['time_ms']
del feature_times['class_label']
plot_feature_barcharts(feature_times)
feature_results = metrics.filter_results(validation_results)
plot_features_f1s(feature_results, datapoint_features[0:1], 1, 1)
plot_features_f1s(feature_results, datapoint_features[1:], 3, 3)
plot_features_f1s(feature_results,
datapoint_features,
5,
2,
plot_type='include')
plot_features_f1s(feature_results,
datapoint_features,