def loading_data(project):
train, test = loading_variable(project +
'_train'), loading_variable(project +
'_test')
dictionary = (loading_variable(project + '_dict_msg'),
loading_variable(project + '_dict_code'))
return train, test, dictionary
plt.hist(data, bins=range(0, 40))
plt.title(check)
plt.xlabel("Length")
plt.ylabel("Frequency")
plt.show()
return mean(data), stdev(data)
elif check == 'Length':
new_data = list()
for d in data:
for f in d:
new_data.append(len(f.split()))
plt.hist(new_data, bins=range(0, 750))
plt.title(check)
plt.xlabel("Length")
plt.ylabel("Frequency")
plt.show()
return mean(new_data), stdev(new_data)
if __name__ == '__main__':
project = 'openstack'
messages, codes = loading_variable(project + '_messages'), loading_variable(project + '_codes')
print(type(messages), type(codes))
mean_, std_ = statistic_msg(data=messages)
print(mean_, std_)
mean_, std_ = statistic_code(data=codes, check='File')
print(mean_, std_)
mean_, std_ = statistic_code(data=codes, check='Length')
print(mean_, std_)
info_label(data=labels_test)
ids_train, ids_test = get_index(data=ids, index=train_index), get_index(
data=ids, index=test_index)
train = (ids_train, labels_train, pad_msg_train, pad_code_train)
test = (ids_test, labels_test, pad_msg_test, pad_code_test)
dict_msg, dict_code = dictionary_commit(
data=pad_msg_train,
type_data='msg'), dictionary_commit(data=pad_code_train,
type_data='code')
return train, test, dict_msg, dict_code
if __name__ == '__main__':
# project = 'openstack'
project = 'qt'
messages, codes = loading_variable(project +
'_messages'), loading_variable(project +
'_codes')
ids, labels = loading_variable(project + '_ids'), convert_label(
loading_variable(project + '_labels'))
info_label(data=labels)
print('Number of instances in commit message %i and commit code %i ' %
(len(messages), len(codes)))
print('Labels: %i' % (len(labels)))
train, test, dict_msg, dict_code = folding_data(pad_msg=messages,
pad_code=codes,
labels=labels,
ids=ids,
n_folds=5)
saving_variable(project + '_train', train)
saving_variable(project + '_test', test)
saving_variable(project + '_dict_msg', dict_msg)