def __init__(self):
self.__db_manager = DBManager()
self.regexp_for_predict_lines = "\d{1,}\s{1,}\d{1}:\w{1,8}.{1,}"
self.__dictionaries_directory = PROJECT_ROOT_DIRECTORY + DICTIONARIES_DIR_NAME
def __init__(self):
self.__db_manager = DBManager()
self.regexp_for_predict_lines = "\d{1,}\s{1,}\d{1}:\w{1,8}.{1,}"
self.__dictionaries_directory = PROJECT_ROOT_DIRECTORY + DICTIONARIES_DIR_NAME
class GeneralHelpers:
def __init__(self):
self.__db_manager = DBManager()
self.regexp_for_predict_lines = "\d{1,}\s{1,}\d{1}:\w{1,8}.{1,}"
self.__dictionaries_directory = PROJECT_ROOT_DIRECTORY + DICTIONARIES_DIR_NAME
def pretty_print_list(self, list_to_print, message):
"""
Pretty prints a given list
:param list_to_print: List, list to print
:param message: String, a message to print before printing list
:return: void
"""
if len(list_to_print):
print(message)
for element in list_to_print:
print(element)
def get_chunks_of_list(self, list_to_chunk, chunk_size):
"""
Returns chunks of new list
:param list_to_chunk: List, List to divide into chunks
:param chunk_size: Int, size of a chunk
:return:
"""
n = max(1, chunk_size)
return [list_to_chunk[i:i + n] for i in range(0, len(list_to_chunk), n)]
def find_root_from_zemberek(self, word):
"""
Finds root of given word from zemberek
:param word: string, word
:return: string, root of word
"""
connection_output = self._make_jar_call(ZEMBEREK_ROOT_FINDER_JAR_FILE_NAME, word)
return connection_output
def correct_misspelling_from_zemberek(self, word):
"""
Corrects misspelled words by asking to zemberek
:param word: string, word to correct
:return: string, corrected word
"""
connection_output = self._make_jar_call(ZEMBEREK_SUGGESTION_FINDER_JAR_FILE_NAME, word)
suggestions = connection_output.split(",")
if len(suggestions):
if word in suggestions:
return word
else:
return suggestions[0]
else:
return word
def save_changes_in_suggestion_cache(self, suggestions_cache):
"""
Saves given suggestion cache to file
:param suggestions_cache: dict, suggestions
:return: void
"""
suggestion_cache_file_path = self.__dictionaries_directory + MODEL_NAME + '/' + SUGGESTION_CACHE_FILE_NAME
self._write_json_to_file(suggestions_cache, suggestion_cache_file_path)
def save_changes_in_root_cache(self, roots_cache):
"""
Saves given roots cache to file
:param roots_cache: roots_cache, dict, roots
:return: void
"""
roots_file_path = self.__dictionaries_directory + MODEL_NAME + '/' + ROOTS_CACHE_FILE_NAME
self._write_json_to_file(roots_cache, roots_file_path)
def load_suggestion_cache(self):
"""
Loads previously asked (to zemberek) suggestion cache
:return: void
"""
suggestion_cache_file_path = self.__dictionaries_directory + MODEL_NAME + '/' + SUGGESTION_CACHE_FILE_NAME
if os.path.isfile(suggestion_cache_file_path):
with open(suggestion_cache_file_path, "r") as suggestions_file:
suggestion_cache = json.load(suggestions_file)
return suggestion_cache
def load_roots_cache(self):
"""
Loads previously asked (to zemberek) word roots cache
:return: void
"""
roots_file_path = self.__dictionaries_directory + MODEL_NAME + '/' + ROOTS_CACHE_FILE_NAME
if os.path.isfile(roots_file_path):
with open(roots_file_path, "r") as roots_file:
roots_cache = json.load(roots_file)
return roots_cache
def generate_arff_file(self, file_path, file_name, arff_data):
"""
Generates arff file
:param file_name: file_name for arff data
:param arff_data: dict, arff_data
:return: string, generated file path
"""
if not os.path.exists(file_path):
os.makedirs(file_path)
arff_file = codecs.open(file_path+file_name, 'w+', encoding='utf-8')
arff.dump(arff_data, arff_file)
arff_file.close()
def generate_random_string(self, n):
"""
Generates random string with size of n
:param n: int, size
:return: string, n-length random string
"""
random_string = ''.join(random.choice('abcdefghijklmnoprstuvyzxw1234567890') for _ in range(n))
return random_string
def generate_random_file_name(self, file_name, extension):
"""
Generates random file name with given file name
:param file_name: string, desired file name root
:return: string, randomized file name
"""
random_file_appendix = self.generate_random_string(5)
final_file_name = file_name + '_' + random_file_appendix + extension
return final_file_name
def get_accuracy_scores_for_years_from_root_dir(self, root_dir):
"""
Returns a dict of __years' classifier scores
:param root_dir: string, path to root directory
:return: dict, __years' classifier scores
"""
years_scores = {}
# Iterating over directories in root directory
for root, dirs, files in os.walk(root_dir):
# Iterating over files in a directory
for file_name in files:
# If it's a txt file we got it. E.g. TTNet_2015_SMO.txt
if file_name.endswith('.txt'):
file_path = root + '/' + file_name
model_name, year, classifier_name = file_name.rstrip('.txt').split("_") #TTNet, 2015, SMO
if not year in years_scores:
years_scores[year] = {}
with open(file_path, 'r') as classifier_log_file:
file_content = classifier_log_file.read()
years_scores[year][classifier_name] = self.get_accuracy_score_from_log_file_content(file_content)
# Calculating mean for each year and sorting
for year, classifiers in years_scores.iteritems():
all_classifier_scores = np.array(classifiers.values())
years_scores[year]['MEAN'] = round(all_classifier_scores.mean(), 2)
years_scores[year] = collections.OrderedDict(sorted(years_scores[year].items()))
sorted_years_scores = collections.OrderedDict(sorted(years_scores.items()))
return sorted_years_scores
def get_accuracy_scores_for_experiment_years_from_root_dir(self, root_dir):
"""
Returns a dict of scores
:param root_dir: string, path to root directory
:return: dict, __years' classifier scores
"""
lines_scores = {}
# Iterating over directories in root directory
for root, dirs, files in os.walk(root_dir):
experiment_scores = []
# Iterating over files in a directory
for file_name in files:
if file_name.endswith('.txt'):
file_path = root + '/' + file_name
with open(file_path, 'r') as classifier_log_file:
file_content = classifier_log_file.read()
accuracy_score_of_log_file = self.get_accuracy_score_from_log_file_content(file_content)
experiment_scores.append(accuracy_score_of_log_file)
if len(experiment_scores):
mean_of_experiment_scores = round(np.array(experiment_scores).mean(), 2)
root_components = root.split("/")
line_name = root_components[-2]
point_name = root_components[-1]
point_name = LINES_DIR_DICT[point_name]
if not line_name in lines_scores:
lines_scores[line_name] = {}
lines_scores[line_name][point_name] = mean_of_experiment_scores
for line, points in lines_scores.iteritems():
lines_scores[line] = collections.OrderedDict(sorted(points.items()))
lines_scores = collections.OrderedDict(sorted(lines_scores.items()))
return lines_scores
def get_accuracy_score_from_log_file_content(self, log_file_content):
"""
Returns accuracy score of log file
:param log_file_content: string
:return: float
"""
regexp_for_accuracy_lines = "Correctly Classified Instances.{1,}"
classifier_accuracy_lines = re.findall(regexp_for_accuracy_lines, log_file_content, re.IGNORECASE)
if len(classifier_accuracy_lines):
accuracy_line = classifier_accuracy_lines[0]
accuracy_line_components = accuracy_line.split() #['Correctly', 'Classified', 'Instances', '209', '41.8', '%']
accuracy = accuracy_line_components[4]
return float(accuracy)
else:
return -1.0
def get_log_files_stats(self, root_dir):
"""
Returns each classifier's monthly accuracy scores from log files.
:param root_dir: string
:return: dict
"""
# Going to log files directory
os.chdir(root_dir)
all_log_files = {}
# iterating over log files of a model (say, Turkcell)
for txt_file in glob("*.txt"):
# openning file
with open(txt_file, 'r') as a_log_file:
# getting log file's name
file_name = a_log_file.name.split(".")[0]
# updating data model with log file's content
all_log_files.update({file_name: a_log_file.read()})
# we got a model and we need it month's lengths (Say, we got 42 tweets for October)
self.model_month_counts = self.__db_manager.get_months_lengths()
all_accuracy_scores = {}
# iterating log files we previously read
for log_file_name, log_file_content in all_log_files.iteritems():
accuracy_scores = self.get_accuracy_scores_per_month_from_log_file(log_file_content)
if not log_file_name in all_accuracy_scores:
all_accuracy_scores[log_file_name] = accuracy_scores
return all_accuracy_scores
def get_accuracy_scores_per_month_from_log_file(self, log_file_content):
"""
Returns each year's [true, total] predictions for given log file content
:param log_file_content: string
:return: dict
"""
predict_lines_in_file = re.findall(self.regexp_for_predict_lines, log_file_content)
correct_vs_total_values = {}
start_idx = 0
# e.g. { 2014: [42, 42, ...., 38] }
for year, months_lengths in self.model_month_counts.iteritems():
# e.g. 42
for month_length in months_lengths:
end_idx = start_idx + month_length
# a months' stats
a_months_stats = [0, month_length] # (Correct, Total)
# slicing months' lines to find lines like: "1 3:positive 3:positive 0 0 *1"
months_lines = predict_lines_in_file[start_idx:end_idx]
# iterating over a month's lines to find accuracy
for month_line in months_lines:
# if line contains + it's an error
if not '+' in month_line:
a_months_stats[0] += 1
if not year in correct_vs_total_values:
correct_vs_total_values[year] = []
correct_vs_total_values[year].append(a_months_stats)
start_idx += month_length
"""
Example correct_vs_total_values:
{
2013: [[25, 43], [26, 42], [21, 42], [31, 44], [25, 42], [24, 40], [27, 42], [23, 42], [23, 43], [28, 41], [25, 42], [21, 39]],
2014: [[35, 55], [19, 39], [25, 46], [32, 43], [23, 42], [25, 42], [27, 42], [28, 45], [25, 45], [27, 45], [23, 42], [15, 27]],
2015: [[25, 43], [32, 56], [34, 57], [36, 56], [31, 54], [35, 55], [37, 68], [38, 57], [27, 51]]
}
"""
# Let's find accuracies now.
accuracy_scores = {}
for year, all_months_predictions in correct_vs_total_values.iteritems():
if not year in accuracy_scores:
accuracy_scores[year] = []
for month_predictions in all_months_predictions:
correct_predictions = month_predictions[0]
total_predictions = month_predictions[1]
one_acc_score = float(correct_predictions) / total_predictions
one_acc_score *= 100
one_acc_score = round(one_acc_score, 2)
accuracy_scores[year].append(one_acc_score)
return accuracy_scores
def find_key_of_given_value_in_dict(self, dictionary, value_to_search):
"""
Returns key of given value
:param dictionary: dict
:param value_to_search: string
:return: string
"""
for k, v in dictionary.iteritems():
if v == value_to_search:
return k
return -1
def find_most_distinct_n_samples(self, document_one, document_two, n_samples):
"""
Returns most distinct n_samples from document_two comparing to document_one (documents are arff data)
:param document_one: list
:param document_two: list
:param n_samples: int
:return: list
"""
pass
def cumulate_years_scores(self, years_scores):
"""
Cumulates and finds averages of all __years' scores
:param years_scores: list
:return: dict
"""
final_result_of_all_experiments = {}
for one_experiments_scores in years_scores:
for line_name, one_lines_scores in one_experiments_scores.iteritems():
if not line_name in final_result_of_all_experiments:
final_result_of_all_experiments[line_name] = {}
for setup_name, score_or_list in one_lines_scores.iteritems():
if not setup_name in final_result_of_all_experiments[line_name]:
final_result_of_all_experiments[line_name][setup_name] = []
final_result_of_all_experiments[line_name][setup_name].append(score_or_list)
for line_name, one_lines_scores in final_result_of_all_experiments.iteritems():
if line_name != "line2":
for setup_name, scores_list in one_lines_scores.iteritems():
final_result_of_all_experiments[line_name][setup_name] = round(np.mean(scores_list), 2)
else:
for setup_name, scores_list in one_lines_scores.iteritems():
np_array_scores_list = np.array(scores_list)
min_mean_max_mean = np.mean(scores_list, axis=0)
final_result_of_all_experiments[line_name][setup_name] = min_mean_max_mean
print('Relative scores:')
pprint.pprint(final_result_of_all_experiments, width=2)
print(final_result_of_all_experiments)
for line_name, score in final_result_of_all_experiments['line3'].iteritems():
target_line1_key = self.get_line1_key_from_line3_key(line_name)
new_line3_score = final_result_of_all_experiments['line1'][target_line1_key] + score
final_result_of_all_experiments['line3'][line_name] = new_line3_score
print('Absolute scores:')
pprint.pprint(final_result_of_all_experiments, width=2)
return final_result_of_all_experiments
def calculate_relative_scores(self, all_scores):
"""
Calculates relative scores of line3 according to line1
:return: void
"""
line1_scores = all_scores['line1']
for setup_name, score in all_scores['line3'].iteritems():
target_line1_key = self.get_line1_key_from_line3_key(setup_name)
difference = score - line1_scores[target_line1_key]
all_scores['line3'][setup_name] = difference
return all_scores
def get_line1_key_from_line3_key(self, line3_key):
"""
Returns relevant line1 key from line3 key
:param line3_key: string
:return: string
"""
regexp_for_target_keys = "(L\d{1}-|\+(2012|2013|2014|2015)_"+ALE_PARTITION_50_KEY+")"
target_line1_key = re.sub(regexp_for_target_keys, "", line3_key)
return target_line1_key
def _write_json_to_file(self, json_data, file_path):
"""
Writes given data to given path
:param data: list or dict, json_data
:param file_path: string
:return:
"""
with open(file_path, 'w') as outfile:
json.dump(json_data, outfile)
def _make_jar_call(self, jar_file_name, word):
"""
Makes a jar call with proper parameters
:param jar_file_name: string, jar file to make call
:param word: string, parameter word
:return: string, connection output
"""
jar_file_path = JAR_FILE_DIR_NAME + jar_file_name
# Making the call
process_call = Popen(['java', '-jar', jar_file_path, word], stdout=PIPE, stderr=STDOUT)
# Getting output
output = process_call.communicate()[0].decode('utf-8')
return output
class GeneralHelpers:
def __init__(self):
self.__db_manager = DBManager()
self.regexp_for_predict_lines = "\d{1,}\s{1,}\d{1}:\w{1,8}.{1,}"
self.__dictionaries_directory = PROJECT_ROOT_DIRECTORY + DICTIONARIES_DIR_NAME
def pretty_print_list(self, list_to_print, message):
"""
Pretty prints a given list
:param list_to_print: List, list to print
:param message: String, a message to print before printing list
:return: void
"""
if len(list_to_print):
print(message)
for element in list_to_print:
print(element)
def get_chunks_of_list(self, list_to_chunk, chunk_size):
"""
Returns chunks of new list
:param list_to_chunk: List, List to divide into chunks
:param chunk_size: Int, size of a chunk
:return:
"""
n = max(1, chunk_size)
return [
list_to_chunk[i:i + n] for i in range(0, len(list_to_chunk), n)
]
def find_root_from_zemberek(self, word):
"""
Finds root of given word from zemberek
:param word: string, word
:return: string, root of word
"""
connection_output = self._make_jar_call(
ZEMBEREK_ROOT_FINDER_JAR_FILE_NAME, word)
return connection_output
def correct_misspelling_from_zemberek(self, word):
"""
Corrects misspelled words by asking to zemberek
:param word: string, word to correct
:return: string, corrected word
"""
connection_output = self._make_jar_call(
ZEMBEREK_SUGGESTION_FINDER_JAR_FILE_NAME, word)
suggestions = connection_output.split(",")
if len(suggestions):
if word in suggestions:
return word
else:
return suggestions[0]
else:
return word
def save_changes_in_suggestion_cache(self, suggestions_cache):
"""
Saves given suggestion cache to file
:param suggestions_cache: dict, suggestions
:return: void
"""
suggestion_cache_file_path = self.__dictionaries_directory + MODEL_NAME + '/' + SUGGESTION_CACHE_FILE_NAME
self._write_json_to_file(suggestions_cache, suggestion_cache_file_path)
def save_changes_in_root_cache(self, roots_cache):
"""
Saves given roots cache to file
:param roots_cache: roots_cache, dict, roots
:return: void
"""
roots_file_path = self.__dictionaries_directory + MODEL_NAME + '/' + ROOTS_CACHE_FILE_NAME
self._write_json_to_file(roots_cache, roots_file_path)
def load_suggestion_cache(self):
"""
Loads previously asked (to zemberek) suggestion cache
:return: void
"""
suggestion_cache_file_path = self.__dictionaries_directory + MODEL_NAME + '/' + SUGGESTION_CACHE_FILE_NAME
if os.path.isfile(suggestion_cache_file_path):
with open(suggestion_cache_file_path, "r") as suggestions_file:
suggestion_cache = json.load(suggestions_file)
return suggestion_cache
def load_roots_cache(self):
"""
Loads previously asked (to zemberek) word roots cache
:return: void
"""
roots_file_path = self.__dictionaries_directory + MODEL_NAME + '/' + ROOTS_CACHE_FILE_NAME
if os.path.isfile(roots_file_path):
with open(roots_file_path, "r") as roots_file:
roots_cache = json.load(roots_file)
return roots_cache
def generate_arff_file(self, file_path, file_name, arff_data):
"""
Generates arff file
:param file_name: file_name for arff data
:param arff_data: dict, arff_data
:return: string, generated file path
"""
if not os.path.exists(file_path):
os.makedirs(file_path)
arff_file = codecs.open(file_path + file_name, 'w+', encoding='utf-8')
arff.dump(arff_data, arff_file)
arff_file.close()
def generate_random_string(self, n):
"""
Generates random string with size of n
:param n: int, size
:return: string, n-length random string
"""
random_string = ''.join(
random.choice('abcdefghijklmnoprstuvyzxw1234567890')
for _ in range(n))
return random_string
def generate_random_file_name(self, file_name, extension):
"""
Generates random file name with given file name
:param file_name: string, desired file name root
:return: string, randomized file name
"""
random_file_appendix = self.generate_random_string(5)
final_file_name = file_name + '_' + random_file_appendix + extension
return final_file_name
def get_accuracy_scores_for_years_from_root_dir(self, root_dir):
"""
Returns a dict of __years' classifier scores
:param root_dir: string, path to root directory
:return: dict, __years' classifier scores
"""
years_scores = {}
# Iterating over directories in root directory
for root, dirs, files in os.walk(root_dir):
# Iterating over files in a directory
for file_name in files:
# If it's a txt file we got it. E.g. TTNet_2015_SMO.txt
if file_name.endswith('.txt'):
file_path = root + '/' + file_name
model_name, year, classifier_name = file_name.rstrip(
'.txt').split("_") #TTNet, 2015, SMO
if not year in years_scores:
years_scores[year] = {}
with open(file_path, 'r') as classifier_log_file:
file_content = classifier_log_file.read()
years_scores[year][
classifier_name] = self.get_accuracy_score_from_log_file_content(
file_content)
# Calculating mean for each year and sorting
for year, classifiers in years_scores.iteritems():
all_classifier_scores = np.array(classifiers.values())
years_scores[year]['MEAN'] = round(all_classifier_scores.mean(), 2)
years_scores[year] = collections.OrderedDict(
sorted(years_scores[year].items()))
sorted_years_scores = collections.OrderedDict(
sorted(years_scores.items()))
return sorted_years_scores
def get_accuracy_scores_for_experiment_years_from_root_dir(self, root_dir):
"""
Returns a dict of scores
:param root_dir: string, path to root directory
:return: dict, __years' classifier scores
"""
lines_scores = {}
# Iterating over directories in root directory
for root, dirs, files in os.walk(root_dir):
experiment_scores = []
# Iterating over files in a directory
for file_name in files:
if file_name.endswith('.txt'):
file_path = root + '/' + file_name
with open(file_path, 'r') as classifier_log_file:
file_content = classifier_log_file.read()
accuracy_score_of_log_file = self.get_accuracy_score_from_log_file_content(
file_content)
experiment_scores.append(accuracy_score_of_log_file)
if len(experiment_scores):
mean_of_experiment_scores = round(
np.array(experiment_scores).mean(), 2)
root_components = root.split("/")
line_name = root_components[-2]
point_name = root_components[-1]
point_name = LINES_DIR_DICT[point_name]
if not line_name in lines_scores:
lines_scores[line_name] = {}
lines_scores[line_name][point_name] = mean_of_experiment_scores
for line, points in lines_scores.iteritems():
lines_scores[line] = collections.OrderedDict(sorted(
points.items()))
lines_scores = collections.OrderedDict(sorted(lines_scores.items()))
return lines_scores
def get_accuracy_score_from_log_file_content(self, log_file_content):
"""
Returns accuracy score of log file
:param log_file_content: string
:return: float
"""
regexp_for_accuracy_lines = "Correctly Classified Instances.{1,}"
classifier_accuracy_lines = re.findall(regexp_for_accuracy_lines,
log_file_content, re.IGNORECASE)
if len(classifier_accuracy_lines):
accuracy_line = classifier_accuracy_lines[0]
accuracy_line_components = accuracy_line.split(
) #['Correctly', 'Classified', 'Instances', '209', '41.8', '%']
accuracy = accuracy_line_components[4]
return float(accuracy)
else:
return -1.0
def get_log_files_stats(self, root_dir):
"""
Returns each classifier's monthly accuracy scores from log files.
:param root_dir: string
:return: dict
"""
# Going to log files directory
os.chdir(root_dir)
all_log_files = {}
# iterating over log files of a model (say, Turkcell)
for txt_file in glob("*.txt"):
# openning file
with open(txt_file, 'r') as a_log_file:
# getting log file's name
file_name = a_log_file.name.split(".")[0]
# updating data model with log file's content
all_log_files.update({file_name: a_log_file.read()})
# we got a model and we need it month's lengths (Say, we got 42 tweets for October)
self.model_month_counts = self.__db_manager.get_months_lengths()
all_accuracy_scores = {}
# iterating log files we previously read
for log_file_name, log_file_content in all_log_files.iteritems():
accuracy_scores = self.get_accuracy_scores_per_month_from_log_file(
log_file_content)
if not log_file_name in all_accuracy_scores:
all_accuracy_scores[log_file_name] = accuracy_scores
return all_accuracy_scores
def get_accuracy_scores_per_month_from_log_file(self, log_file_content):
"""
Returns each year's [true, total] predictions for given log file content
:param log_file_content: string
:return: dict
"""
predict_lines_in_file = re.findall(self.regexp_for_predict_lines,
log_file_content)
correct_vs_total_values = {}
start_idx = 0
# e.g. { 2014: [42, 42, ...., 38] }
for year, months_lengths in self.model_month_counts.iteritems():
# e.g. 42
for month_length in months_lengths:
end_idx = start_idx + month_length
# a months' stats
a_months_stats = [0, month_length] # (Correct, Total)
# slicing months' lines to find lines like: "1 3:positive 3:positive 0 0 *1"
months_lines = predict_lines_in_file[start_idx:end_idx]
# iterating over a month's lines to find accuracy
for month_line in months_lines:
# if line contains + it's an error
if not '+' in month_line:
a_months_stats[0] += 1
if not year in correct_vs_total_values:
correct_vs_total_values[year] = []
correct_vs_total_values[year].append(a_months_stats)
start_idx += month_length
"""
Example correct_vs_total_values:
{
2013: [[25, 43], [26, 42], [21, 42], [31, 44], [25, 42], [24, 40], [27, 42], [23, 42], [23, 43], [28, 41], [25, 42], [21, 39]],
2014: [[35, 55], [19, 39], [25, 46], [32, 43], [23, 42], [25, 42], [27, 42], [28, 45], [25, 45], [27, 45], [23, 42], [15, 27]],
2015: [[25, 43], [32, 56], [34, 57], [36, 56], [31, 54], [35, 55], [37, 68], [38, 57], [27, 51]]
}
"""
# Let's find accuracies now.
accuracy_scores = {}
for year, all_months_predictions in correct_vs_total_values.iteritems(
):
if not year in accuracy_scores:
accuracy_scores[year] = []
for month_predictions in all_months_predictions:
correct_predictions = month_predictions[0]
total_predictions = month_predictions[1]
one_acc_score = float(correct_predictions) / total_predictions
one_acc_score *= 100
one_acc_score = round(one_acc_score, 2)
accuracy_scores[year].append(one_acc_score)
return accuracy_scores
def find_key_of_given_value_in_dict(self, dictionary, value_to_search):
"""
Returns key of given value
:param dictionary: dict
:param value_to_search: string
:return: string
"""
for k, v in dictionary.iteritems():
if v == value_to_search:
return k
return -1
def find_most_distinct_n_samples(self, document_one, document_two,
n_samples):
"""
Returns most distinct n_samples from document_two comparing to document_one (documents are arff data)
:param document_one: list
:param document_two: list
:param n_samples: int
:return: list
"""
pass
def cumulate_years_scores(self, years_scores):
"""
Cumulates and finds averages of all __years' scores
:param years_scores: list
:return: dict
"""
final_result_of_all_experiments = {}
for one_experiments_scores in years_scores:
for line_name, one_lines_scores in one_experiments_scores.iteritems(
):
if not line_name in final_result_of_all_experiments:
final_result_of_all_experiments[line_name] = {}
for setup_name, score_or_list in one_lines_scores.iteritems():
if not setup_name in final_result_of_all_experiments[
line_name]:
final_result_of_all_experiments[line_name][
setup_name] = []
final_result_of_all_experiments[line_name][
setup_name].append(score_or_list)
for line_name, one_lines_scores in final_result_of_all_experiments.iteritems(
):
if line_name != "line2":
for setup_name, scores_list in one_lines_scores.iteritems():
final_result_of_all_experiments[line_name][
setup_name] = round(np.mean(scores_list), 2)
else:
for setup_name, scores_list in one_lines_scores.iteritems():
np_array_scores_list = np.array(scores_list)
min_mean_max_mean = np.mean(scores_list, axis=0)
final_result_of_all_experiments[line_name][
setup_name] = min_mean_max_mean
print('Relative scores:')
pprint.pprint(final_result_of_all_experiments, width=2)
print(final_result_of_all_experiments)
for line_name, score in final_result_of_all_experiments[
'line3'].iteritems():
target_line1_key = self.get_line1_key_from_line3_key(line_name)
new_line3_score = final_result_of_all_experiments['line1'][
target_line1_key] + score
final_result_of_all_experiments['line3'][
line_name] = new_line3_score
print('Absolute scores:')
pprint.pprint(final_result_of_all_experiments, width=2)
return final_result_of_all_experiments
def calculate_relative_scores(self, all_scores):
"""
Calculates relative scores of line3 according to line1
:return: void
"""
line1_scores = all_scores['line1']
for setup_name, score in all_scores['line3'].iteritems():
target_line1_key = self.get_line1_key_from_line3_key(setup_name)
difference = score - line1_scores[target_line1_key]
all_scores['line3'][setup_name] = difference
return all_scores
def get_line1_key_from_line3_key(self, line3_key):
"""
Returns relevant line1 key from line3 key
:param line3_key: string
:return: string
"""
regexp_for_target_keys = "(L\d{1}-|\+(2012|2013|2014|2015)_" + ALE_PARTITION_50_KEY + ")"
target_line1_key = re.sub(regexp_for_target_keys, "", line3_key)
return target_line1_key
def _write_json_to_file(self, json_data, file_path):
"""
Writes given data to given path
:param data: list or dict, json_data
:param file_path: string
:return:
"""
with open(file_path, 'w') as outfile:
json.dump(json_data, outfile)
def _make_jar_call(self, jar_file_name, word):
"""
Makes a jar call with proper parameters
:param jar_file_name: string, jar file to make call
:param word: string, parameter word
:return: string, connection output
"""
jar_file_path = JAR_FILE_DIR_NAME + jar_file_name
# Making the call
process_call = Popen(['java', '-jar', jar_file_path, word],
stdout=PIPE,
stderr=STDOUT)
# Getting output
output = process_call.communicate()[0].decode('utf-8')
return output
