from features_calculator import featuresCalculator
from learning.TagsLoader import TagsLoader
import main_manager as mm
import numpy as np
currentDirectory = str(os.getcwd())
if __name__ == "__main__":
np.random.seed(123)
#step 1: features calculate
wdir = os.getcwd()
file_in = str(wdir) + r'/../data/directed/DIP/input/DIP.txt'
output_dir = str(wdir) + r'/../data/directed/DIP/features/take_connected'
calculator = featuresCalculator()
features_list = featuresList.featuresList(
directed=True, analysisType='nodes').getFeatures()
features_list.remove('kcore')
features_list.remove('flow')
features_list.remove('ab')
features_list.remove('motif4')
result = calculator.calculateFeatures(features_list,
file_in,
output_dir,
directed=True,
analysisType='nodes')
# step 2: learning phase
all_MF_classifications = [
'0005326', '0005261', '0019215', '0004016', '0001076', '0019911',
'0003913', '0016866', '0004812', '0005544', '0003712', '0000166',
'0016740', '0004715', '0016746', '0004896', '0005319', '0004383',
from learning.TagsLoader import TagsLoader
import main_manager as mm
currentDirectory = str(os.getcwd())
if __name__ == "__main__":
#step 1: features calculate
wdir = os.getcwd()
file_in = str(
wdir) + r'/../data/directed/social_sign/slashdot/input/slashdot.txt'
output_dir = str(wdir) + r'/../data/directed/social_sign/slashdot/features'
calculator = featuresCalculator()
features_list = featuresList.featuresList(True, 'edges').getFeatures()
features_list.remove('motif4')
features_list.remove('flow')
features_list.remove('ab')
features_list.remove('hierarchy_energy')
features_list.remove('edge_flow')
features_list.remove('edge_betweenness')
print(features_list)
result = calculator.calculateFeatures(features_list,
file_in,
output_dir,
True,
'edges',
parallel=False)
print(result[1].keys())
print(len(result[1]))
if __name__ == "__main__":
wdir = os.getcwd()
for i in range(1,2):
if(i<10):
snap = '000'+str(i)
else:
snap = '00' + str(i)
print (snap)
# snap = '0001'
file_in = str(wdir) + r'/../data/directed/live_journal/'+snap+r'/input/graph.txt'
output_dir = str(wdir) + r'/../data/directed/live_journal/'+snap+r'/features'
calculator = featuresCalculator()
features_list = featuresList.featuresList(True, 'nodes').getFeatures()
features_list.remove('flow')
features_list.remove('ab')
features_list.remove('eccentricity')
features_list.remove('load_centrality')
features_list.remove('hierarchy_energy')
result = calculator.calculateFeatures(features_list, file_in, output_dir, True, 'nodes')
motif_path = str(wdir) + r'/../graph-fetures/algo/motifVariations'
result = features.calc_fetures_vertices(file_in, motif_path, output_dir, directed=True, takeConnected=True, fetures_list=directed_lj, return_map=True)
print (result[1].keys())
doi = ['summer'
, 'the beatles'
, 'cars'
def calc_by_train_size(graph_name, test_sizes, f_output, random_state, load,
deep):
output_result_dir = {}
for test_size in test_sizes:
output_result_dir[
test_size] = './../data/result_social_sign/' + graph_name + '/' + str(
test_size) + '/'
if (not os.path.exists(output_result_dir[test_size])):
os.mkdir(output_result_dir[test_size])
if not load:
wdir = os.getcwd()
file_in = str(
wdir
) + r'/../data/directed/social_sign/' + graph_name + '/input/' + graph_name + '.txt'
classification_result = [graph_name + '-tags']
directory_tags_path = str(
wdir) + r'/../data/directed/social_sign/' + graph_name + '/tags/'
# file_in = str(wdir) + r'/../data/directed/social_sign/epinions/input/epinions.txt'
# classification_wiki_result = ['epinions-tags']
# directory_tags_path = str(wdir) + r'/../data/directed/social_sign/epinions/tags/'
# file_in = str(wdir) + r'/../data/roi-graph.txt'
# classification_wiki_result = ['roi-tags']
# directory_tags_path = str(wdir) + r'/../data/'
print(' start reload graph')
# [ggt, gnx] = initGraph.init_graph(draw = False);
gnx = initGraph.init_graph(draw=False,
file_name=file_in,
directed=True,
Connected=True)
print(' finish reload graph')
calculator = featuresCalculator()
features_list = featuresList.featuresList(True, 'edges').getFeatures()
features_list.remove('flow')
features_list.remove('ab')
features_list.remove('hierarchy_energy')
features_list.remove('edge_flow')
features_list.remove('edge_betweenness')
features_list.remove('motif4')
if (graph_name in ['epinions']):
features_list.remove('kcore')
print(features_list)
# output_dir = str(wdir) + r'/../data/directed/social_sign/epinions/features'
output_dir = str(
wdir
) + r'/../data/directed/social_sign/' + graph_name + '/features'
# output_dir = str(wdir) + r'/../data/'
result = calculator.calculateFeatures(features_list,
file_in,
output_dir,
True,
'edges',
parallel=False)
tagsLoader = TagsLoader(directory_tags_path, classification_result)
tagsLoader.load_edges()
# tags = tagsLoader.calssification_to_edge_to_tag['epinions-tags']
tags = tagsLoader.calssification_to_edge_to_tag[
classification_result[0]]
# tags = tagsLoader.calssification_to_edge_to_tag['roi-tags']
print(tagsLoader.calssification_to_edge_to_tag.keys())
remove_sign_zero(gnx, tags)
edges = gnx.edges()
print(len(edges))
print(len(gnx.nodes()))
print(len(tags))
s_e = set(edges)
s_t = set(tags.keys())
print(s_e.difference(s_t))
print(s_t.difference(s_e))
new_tags = []
for e in edges:
new_tags.append(tags[e])
# random_state = random.randint(0,len(edges))
# random_state = 2
X_train = {}
X_test = {}
Y_train = {}
Y_test = {}
for test_size in test_sizes:
x_train, x_test, y_train, y_test = train_test_split(
edges,
new_tags,
test_size=test_size,
random_state=random_state)
X_train[test_size] = x_train
X_test[test_size] = x_test
Y_train[test_size] = y_train
Y_test[test_size] = y_test
# print X_train
# print X_test
# print result[1]
train_features, test_features = calc_local_sign_features(
gnx, X_train, tags, result[1])
# print train_features
# print test_features
#
for test_size in train_features.keys():
with open(
output_result_dir[test_size] + 'train_features_' +
graph_name + '.dump', 'wb') as f:
pickle.dump(train_features[test_size], f)
with open(
output_result_dir[test_size] + 'test_features_' +
graph_name + '.dump', 'wb') as f:
pickle.dump(test_features[test_size], f)
#
else:
for test_size in test_sizes:
print(output_result_dir[test_size] + '/train_features_' +
graph_name + '.dump')
with open(
output_result_dir[test_size] + '/train_features_' +
graph_name + '.dump', 'rb') as f:
train_features_specific = pickle.load(f)
with open(
output_result_dir[test_size] + '/test_features_' +
graph_name + '.dump', 'rb') as f:
test_features_specific = pickle.load(f)
print(test_size)
print(graph_name)
f_output.writelines(str(test_size) + '\n')
# train_features_specific = train_features[test_size]
# test_features_specific = test_features[test_size]
print('local')
f_output.writelines('local\n')
perform_learning(train_features_specific,
test_features_specific,
f_output,
local=True,
gglobal=False,
deep=deep)
print('global')
f_output.writelines('global\n')
perform_learning(train_features_specific,
test_features_specific,
f_output,
local=False,
gglobal=True,
deep=deep)
print('Both')
f_output.writelines('Both\n')
perform_learning(train_features_specific,
test_features_specific,
f_output,
local=True,
gglobal=True,
deep=deep)
f_output.flush()