def adamic_adar_scores(g_train, train_test_split):
if g_train.is_directed(): # Only works for undirected graphs
g_train = g_train.to_undirected()
adj_train, train_edges, train_edges_false, val_edges, val_edges_false, \
test_edges, test_edges_false = train_test_split # Unpack input
start_time = time.time()
aa_scores = {}
# Calculate scores
aa_matrix = np.zeros(adj_train.shape)
for u, v, p in nx.adamic_adar_index(g_train, ebunch=get_ebunch(train_test_split)): # (u, v) = node indices, p = Adamic-Adar index
aa_matrix[u][v] = p
aa_matrix[v][u] = p # make sure it's symmetric
aa_matrix = aa_matrix / aa_matrix.max() # Normalize matrix
runtime = time.time() - start_time
aa_roc, aa_ap = get_roc_score(test_edges, test_edges_false, aa_matrix)
aa_scores['test_roc'] = aa_roc
# aa_scores['test_roc_curve'] = aa_roc_curve
aa_scores['test_ap'] = aa_ap
aa_scores['runtime'] = runtime
return aa_scores
def getaa(i):
t = time.time()
l = list(nx.adamic_adar_index(g, E[i:i + step]))
t = time.time() - t
print(' finished edges [{: 7d}, {: 7d}) in {: 7.2f} s\n'
.format(i, i + step, t), end='')
sys.stdout.flush()
return l
def train_feature(fn_list, input_file='../../output/svm_feature_merge/training.txt(split)'):
percentage_of_initial_adopter = 0.1
g = load_graph('../../data/graph.txt')
b = load_json('../../data/Business.txt')
b = {b['business_id'][i]: ((b['latitude'][i], b['longitude'][i]), b['stars'][i]) for i in range(len(b['business_id']))}
u_location = load_user_location('../../data/user_location.txt')
t = load_idea(input_file)
for fn in fn_list:
fn = '../../output/svm_feature_merge/' + fn
d = __load(fn)
with open(fn+'(f0).txt', 'w') as f0:
with open(fn+'(f1).txt', 'w') as f1:
with open(fn+'(f2).txt', 'w') as f2:
with open(fn+'(f3).txt', 'w') as f3:
for k, v in d.items():
print k
user = [q for q, qq in v]
label = [qq for q, qq in v]
n = get_node_by_idea2(t, k)
print len(n['date'])
index = sorted(range(len(n['date'])), key=lambda k: n['date'][k])
# level = [n['level'][i] for i in index]
node = [n['node'][i] for i in index]
initial_adopters = node[: int(percentage_of_initial_adopter*len(node))]
assert(len(set(user) & set(initial_adopters)) == 0)
# b_avg_stars = np.average([level[: 0.1*len(level)]])
for e, lb in zip(user,label):
features = []
# Distance with business
features.append(__distance(b[k][0], u_location[e]))
# Difference between user average stars and business average stars by adopters
# Average distance with initial adopters
features.append(np.average([__distance(u_location[a], u_location[e]) for a in initial_adopters]))
# Percentage of friends in initial adopters
features.append(float(sum([1 if g.has_edge(e, a) else 0 for a in initial_adopters]))/len(initial_adopters))
# Average adar
preds = nx.adamic_adar_index(g, [(e, a) for a in initial_adopters if g.has_node(a) and g.has_node(e)])
try:
preds = [p for u, v, p in preds]
preds = preds if preds else 0
except Exception:
preds = 0
features.append(np.average(preds))
f0.write('{0} {1} {2} {3}\n'.format(k, e, features[0], lb))
f1.write('{0} {1} {2} {3}\n'.format(k, e, features[1], lb))
f2.write('{0} {1} {2} {3}\n'.format(k, e, features[2], lb))
f3.write('{0} {1} {2} {3}\n'.format(k, e, features[3], lb))
def predictLinksAdamicAdar(nodesAtHop, itemNodeIds, userNodeIds, directory, item):
scores = {}
GCombined = nx.read_edgelist(directory + 'Edge_List_Combined_' + item + '.txt')
preds = nx.adamic_adar_index(GCombined)
for u, v, p in preds:
if not u in scores:
scores[u] = {}
scores[u][v] = p
with open(directory + 'AdamicAdar', 'wb') as outfile:
pickle.dump(scores, outfile)
def graph_stats(distance_couple, net):
distances = []
common_neighbors = []
jaccard = []
adamic = []
edge_bet = []
edge_betweeness = nx.edge_betweenness_centrality(net)
for couple in distance_couple:
distances.append(couple[1])
common_neighbors.append(len(list(nx.common_neighbors(net, couple[0][0], couple[0][1]))))
jaccard.append(list(nx.jaccard_coefficient(net, [(couple[0][0], couple[0][1])]))[0][2])
adamic.append(list(nx.adamic_adar_index(net, [(couple[0][0], couple[0][1])]))[0][2])
try:
edge_bet.append(edge_betweeness[couple[0]])
except KeyError:
edge_bet.append(edge_betweeness[(couple[0][1], couple[0][0])])
r_dist = 10.0/max(distances)
r_n = 10.0/max(common_neighbors)
r_j = 10.0/max(jaccard)
r_a = 10.0/max(adamic)
r_e = 10.0/max(edge_bet)
distances = [j * r_dist for j in distances]
common_neighbors = [j * r_n for j in common_neighbors]
jaccard = [j * r_j for j in jaccard]
adamic = [j * r_a for j in adamic]
edge_bet = [j * r_e for j in edge_bet]
plt.loglog(common_neighbors, color='b', label='common_neighbors')
plt.loglog(distances, color='r', label='distances')
plt.savefig('node_similarity/stats_cm.png', format='png')
plt.close()
plt.loglog(jaccard, color='b', label='jaccard')
plt.loglog(distances, color='r', label='distances')
plt.savefig('node_similarity/stats_j.png', format='png')
plt.close()
plt.loglog(adamic, color='b', label='adamic')
plt.loglog(distances, color='r', label='distances')
plt.savefig('node_similarity/stats_aa.png', format='png')
plt.close()
plt.loglog(edge_bet, color='b', label='edge betwenness')
plt.loglog(distances, color='r', label='distances')
plt.savefig('node_similarity/stats_eb.png', format='png')
plt.close()
def train_feature(input_file='../../data/training.txt(split)', output='../../output/svm_training.txt'):
percentage_of_initial_adopter = 0.1
g = load_graph('../../data/graph.txt')
b = load_json('../../data/Business.txt')
b = {b['business_id'][i]: ((b['latitude'][i], b['longitude'][i]), b['stars'][i]) for i in range(len(b['business_id']))}
u_location = load_user_location('../../data/user_location.txt')
t = load_idea(input_file)
ideas = list(set(t['idea']))
l = None
for m in range(len(ideas)):
print m
n = get_node_by_idea2(t, ideas[m])
index = sorted(range(len(n['date'])), key=lambda k: n['date'][k])
# level = [n['level'][i] for i in index]
node = [n['node'][i] for i in index]
initial_adopters = node[: int(percentage_of_initial_adopter*len(node))]
laters = node[int(percentage_of_initial_adopter*len(node)):]
# b_avg_stars = np.average([level[: 0.1*len(level)]])
for e in laters:
features = []
# Distance with business
features.append(__distance(b[ideas[m]][0], u_location[e]))
# Difference between user average stars and business average stars by adopters
# Average distance with initial adopters
features.append(np.average([__distance(u_location[a], u_location[e]) for a in initial_adopters]))
# Percentage of friends in initial adopters
features.append(float(sum([1 if g.has_edge(e, a) else 0 for a in initial_adopters]))/len(initial_adopters))
# Average adar
preds = nx.adamic_adar_index(g, [(e, a) for a in initial_adopters if g.has_node(a) and g.has_node(e)])
try:
preds = [p for u, v, p in preds]
preds = preds if preds else 0
except Exception:
preds = 0
features.append(np.average(preds))
features = np.array([features])
l = features if l is None else np.concatenate((l, features))
dump_svmlight_file(l, [1]*l.shape[0], output)
def test_feature(input='../../data/test_data/test_data_q1.txt', output='../../output/svm_testing_q1_'):
g = load_graph('../../data/graph.txt')
b = load_json('../../data/Business.txt')
b = {b['business_id'][i]: ((b['latitude'][i], b['longitude'][i]), b['stars'][i]) for i in range(len(b['business_id']))}
u_location = load_user_location('../../data/user_location.txt')
t = load_idea('../../data/testing.txt')
with open('../../data/testing_business.txt', 'r') as f:
tb = f.read().strip().split()
with open(input, 'r') as f:
test = [[e for e in l.strip().split()] for l in f]
for i, (business, initial_adopters) in enumerate(zip(tb, test)):
print i
l = None
answers = []
n = get_node_by_idea2(t, business)
ans = set(n['node'])-set(initial_adopters)
candidates = set(g.nodes())-set(initial_adopters)
for e in candidates:
features = []
# Distance with business
features.append(__distance(b[business][0], u_location[e]))
# Difference between user average stars and business average stars by adopters
# Average distance with initial adopters
features.append(np.average([__distance(u_location[a], u_location[e]) for a in initial_adopters]))
# Percentage of friends in initial adopters
features.append(float(sum([1 if g.has_edge(e, a) else 0 for a in initial_adopters]))/len(initial_adopters))
# Average adar
preds = nx.adamic_adar_index(g, [(e, a) for a in initial_adopters if g.has_node(a) and g.has_node(e)])
try:
preds = [p for u, v, p in preds]
preds = preds if preds else 0
except Exception:
preds = 0
features.append(np.average(preds))
features = np.array([features])
l = features if l is None else np.concatenate((l, features))
answers.append(1 if e in ans else 0)
dump_svmlight_file(l, answers, output+str(i)+'.txt')
def calculate_similarities(E1, tmp_string=None):
features = []
ytarget = []
tmp_feat = {}
tmp_y = {}
temp_common_neighbors = []
node_list = list(E1.nodes)
for node1 in node_list:
for node2 in node_list:
connected = 0
if tmp_string == "n":
if nodes_connected(E1, node1, node2):
continue
else:
temp = nx.jaccard_coefficient(E1, [(node1, node2)])
for u, v, p in temp:
tmp_feat[u, v] = [p]
tmp_y[u, v] = connected
temp = nx.adamic_adar_index(E1, [(node1, node2)])
try:
for u, v, p in temp:
tmp_feat[u, v].append(p)
tmp_y[u, v] = connected
except:
tmp_feat[u, v].append(0.0)
tmp_y[u, v] = connected
temp = nx.preferential_attachment(E1, [(node1, node2)])
for u, v, p in temp:
tmp_feat[u, v].append(p)
tmp_y[u, v] = connected
temp = sorted(nx.common_neighbors(E1, node1, node2))
temp_common_neighbors = []
temp = sorted(nx.common_neighbors(E1, node1, node2))
for common_neighbor in temp:
temp_common_neighbors.append(common_neighbor)
tmp_feat[node1, node2].append(len(temp_common_neighbors))
#Ill use Exception Checking cause we have DiGraph and maybe there is no path between to nodes
#so an error would compile
try:
length = nx.shortest_path_length(E1, node1, node2)
tmp_feat[node1, node2].append(length)
tmp_y[node1, node2] = connected
except:
tmp_feat[node1, node2].append(0.0)
tmp_y[node1, node2] = connected
else:
if nodes_connected(E1, node1, node2):
connected = 1
temp = nx.jaccard_coefficient(E1, [(node1, node2)])
for u, v, p in temp:
tmp_feat[u, v] = [p]
tmp_y[u, v] = connected
temp = nx.adamic_adar_index(E1, [(node1, node2)])
try:
for u, v, p in temp:
tmp_feat[u, v].append(p)
tmp_y[u, v] = connected
except:
tmp_feat[u, v].append(0.0)
tmp_y[u, v] = connected
temp = nx.preferential_attachment(E1, [(node1, node2)])
for u, v, p in temp:
tmp_feat[u, v].append(p)
tmp_y[u, v] = connected
temp = sorted(nx.common_neighbors(E1, node1, node2))
temp_common_neighbors = []
temp = sorted(nx.common_neighbors(E1, node1, node2))
for common_neighbor in temp:
temp_common_neighbors.append(common_neighbor)
tmp_feat[node1, node2].append(len(temp_common_neighbors))
#Ill use Exception Checking cause we have DiGraph and maybe there is no path between to nodes
#so an error would compile
try:
length = nx.shortest_path_length(E1, node1, node2)
tmp_feat[node1, node2].append(length)
tmp_y[node1, node2] = connected
except:
tmp_feat[node1, node2].append(0.0)
tmp_y[node1, node2] = connected
for key, value in tmp_feat.items():
features.append(value)
for key, value in tmp_y.items():
ytarget.append(value)
features_numpy = numpy.array(features)
y_target_numpy = numpy.array(ytarget)
return features_numpy, y_target_numpy
G, [(i, j)])) # do >0
# print(prediction2)
# print(prediction2[0][2])
# prediction2 = prediction2[0][2]
# print('Model2 Pred:' + str(prediction2))
########## Jaccard Coefficient ##########
prediction3 = sorted(nx.jaccard_coefficient(G,
[(i, j)])) # do >0
# print(prediction3)
# print(prediction3[0][2])
# prediction3 = prediction3[0][2]
# print('Model3 Pred:' + str(prediction3))
########## Adamic Adar Index ##########
prediction4 = sorted(nx.adamic_adar_index(G,
[(i, j)])) # do >0
# print(prediction4)
# print(prediction4[0][2])
# prediction4 = prediction4[0][2]
# print('Model4 Pred:' + str(prediction4))
########## Prederential Attachment ##########
prediction5 = sorted(nx.preferential_attachment(
G, [(i, j)])) # do >0
# print(prediction5)
# print(prediction5[0][2])
# prediction5 = prediction5[0][2]
# print('Model5 Pred:' + str(prediction5))
########## Tried couple different ways to get the score up ##########
########## Really thought something like this would work well ##########
def predict(self, node_pairs):
predictions = adamic_adar_index(self.graph, node_pairs)
return list(predictions)
def link_prediction(G, query_nodes, target_nodes, n_edges, start_dist, alg = "ra"):
"""Selects a random set of links between based on the scores calculated by
a standard link-prediction algorithm from networkx library
Parameters
----------
G : Networkx graph
The graph from which the team will be selected.
query : list
The set of nodes from which random walker starts.
target : list
The set of nodes from where the random walker ends.
n_edges : integer
the number of links to be added
start_dist: list
The starting distribution over the query set
alg: string
A string describing the link-prediction algorithm to be used
Returns
-------
links : list
The set of links that reduce the absorbing RW centrality
ac_scores: list
The set of scores of adding the links
"""
assert alg in ["ra", "pa", "jaccard", "aa"], "alg must be one of [\"ra\", \"pa\", \"jaccard\", \"aa\"]."
H = G.copy()
query_set_size = len(query_nodes)
map_query_to_org = dict(zip(query_nodes, range(query_set_size)))
P = csc_matrix(nx.google_matrix(H, alpha=1))
P_abs = P[list(query_nodes),:][:,list(query_nodes)]
F = compute_fundamental(P_abs)
row_sums = start_dist.dot(F.sum())[0,0]
candidates = list(product(query_nodes, target_nodes))
eligible = [candidates[i] for i in range(len(candidates))
if H.has_edge(candidates[i][0], candidates[i][1]) == False]
links_to_add = []
if alg == 'ra':
preds = nx.resource_allocation_index(H, eligible)
elif alg == 'jaccard':
preds = nx.jaccard_coefficient(H, eligible)
elif alg == 'aa':
preds = nx.adamic_adar_index(H, eligible)
elif alg == 'pa':
preds = nx.preferential_attachment(H, eligible)
for u,v,p in preds:
links_to_add.append((u,v,p))
links_to_add.sort(key=lambda x: x[2], reverse = True)
ac_scores = []
ac_scores.append(row_sums)
i = 0
while i < n_edges:
F_updated = update_fundamental_mat(F, H, map_query_to_org, links_to_add[i][0])
H.add_edge(links_to_add[i][0], links_to_add[i][1])
abs_cen = start_dist.dot(F_updated.sum(axis = 1))[0,0]
F = F_updated
ac_scores.append(abs_cen)
i += 1
return links_to_add, ac_scores
print('Reading %s_topological_network.csv...' %
(prog_languages[prog_lang_id]))
t_network = []
for row in data:
dev_id_1 = int(row[0])
dev_id_2 = int(row[1])
t_network.append((dev_id_1, dev_id_2))
csvfile.close()
with open('../Files/topological_metrics.csv', 'a') as a:
metrics_file = csv.writer(a, delimiter=',')
print('Writing topological metrics for', prog_languages[prog_lang_id])
for dev_pair in t_network:
neighborhood_overlap = nx.jaccard_coefficient(G, [dev_pair])
adamic_acar = nx.adamic_adar_index(G, [dev_pair])
preferential_attachment = nx.preferential_attachment(G, [dev_pair])
for u, v, p in neighborhood_overlap:
NO = p
for u, v, p in adamic_acar:
AA = p
for u, v, p in preferential_attachment:
PA = p
metrics_file.writerow(
[prog_lang_id, dev_pair[0], dev_pair[1], NO, AA, PA])
a.close()
def calc_indexes(G, test_edges, aa=False):
"""
Input -
G - is the graph based upon which we calculate the indexes
test_edges - are the edges for which we calculate the index
aa - whether to calculate the adamic adder index or not
Calculates all the indixes of the graph for the test_edgessuch as
1) Common Neighbors
2) Jaccard
3) Preferential Attachment
4) Adamic Adder
"""
#All indexes stored as a dictionary
indexes = {}
#Initiazing
jaccard_arr = []
common_arr = []
preferential_arr = []
adamic_adder = []
aa1 = nx.adamic_adar_index(G, test_edges)
jc = nx.jaccard_coefficient(G, test_edges)
pa = nx.preferential_attachment(G, test_edges)
for edge in test_edges:
#Loading the nodes
node1 = edge[0]
node2 = edge[1]
node_list = [node1, node2]
#Neighbors of the nodes
node1_neighbors = set(list(G.neighbors(node1)))
node2_neighbors = set(list(G.neighbors(node2)))
#Union of the neighbors
union_neighbors = list(node1_neighbors.union(node2_neighbors))
#Intersection of the neighbors
intersection_neighbors = list(
node1_neighbors.intersection(node2_neighbors))
#Jaccard Index
jaccard_index = 0
if len(union_neighbors) != 0:
jaccard_index = len(intersection_neighbors) / len(union_neighbors)
#Common Neighbors
common_neighbors = len(intersection_neighbors)
#Preferential Attachment
preferential_attachment = len(node1_neighbors) * len(node2_neighbors)
#Appending the indexes into the arrays
jaccard_arr.append(jaccard_index)
common_arr.append(common_neighbors)
preferential_arr.append(preferential_attachment)
#if adamic_adder is to be calculated
if aa == True:
adamic_add = 0
for neighbor in intersection_neighbors:
nfneighbors_of_neighbors = len(list(G.neighbors(neighbor)))
#Adamic adder doesn't make sense for nfneighbors_of_neighbors == 0 or 1
if nfneighbors_of_neighbors != 1 and nfneighbors_of_neighbors != 0:
aa_score = 1 / np.log(nfneighbors_of_neighbors)
adamic_add += aa_score
adamic_adder.append(adamic_add)
indexes['jaccard_index'] = jaccard_arr
indexes['common_neighbors'] = common_arr
indexes['preferential_attachment'] = preferential_arr
indexes['adamic_adder'] = adamic_adder
# print(adamic_adder[:10],aa[:10])
# print(jaccard_arr[:10],jc[:10])
# print(preferential_arr[:10],pa[:10])
#print(indexes['adamic_adder'])
#print(adamic_adder)
#print(jaccard_arr)
# for u, v, p in jc:
# print(u, v, p)
# ind = test_edges.index((u,v))
# print(jaccard_arr[ind],ind)
# for u, v, p in pa:
# print(u, v, p)
# ind = test_edges.index((u,v))
# print(preferential_arr[ind],ind)
# for u, v, p in aa1:
# print(u, v, p)
# ind = test_edges.index((u,v))
# print(adamic_adder[ind],ind)
return indexes
auth2.append(b)
length = len(auth1)
print length
for i in range(0, length):
for j in range(0, length):
lab = 0
if (auth1[i] != auth2[j]):
fet = ''
d = 0
f = 0
pred1 = nx.adamic_adar_index(g, [(auth1[i], auth2[j])])
pred2 = nx.jaccard_coefficient(g, [(auth1[i], auth2[j])])
pred3 = nx.preferential_attachment(g, [(auth1[i], auth2[j])])
for item in pred1:
x, y, z = item
if (z == 0.0):
d = 1
fet = fet + str(z) + ' '
#print x + ' '+ y +' ' + str(z)
for item in pred2:
x, y, z = item
if (z == 0.0):
f = 1
fet = fet + str(z) + ' '
def get_feature(source, target):
features = {}
def set_feature(name, val):
if name not in features:
features[name] = val
def cosine_distance(node_list1, node_list2):
id2index = dict([
(id, i) for i, id in enumerate(set(node_list1 + node_list2))
])
a = np.zeros((len(id2index), ))
b = np.zeros((len(id2index), ))
for key in node_list1:
a[id2index[key]] = 1
for key in node_list2:
b[id2index[key]] = 1
#return distance.cosine(a, b)
try:
source_succ = set(digraph.successors(source))
source_pred = set(digraph.predecessors(source))
target_succ = set(digraph.successors(target))
target_pred = set(digraph.predecessors(target))
set_feature('len_source_successors', len(source_succ))
set_feature('len_target_successors', len(target_succ))
set_feature('len_source_predecessors', len(source_pred))
set_feature('len_target_predecessors', len(target_pred))
common_succ = len(source_succ.intersection(target_succ))
common_pred = len(source_pred.intersection(source_pred))
set_feature('common_successor_number', common_succ)
set_feature('common_predecessor_number', common_pred)
succ_union = source_succ.union(target_succ)
pred_union = source_pred.union(target_pred)
set_feature(
'jaccard_distance_between_successors',
common_succ / len(succ_union) if len(succ_union) != 0 else 0)
set_feature(
'jaccard_distance_between_predecessors',
common_pred / len(pred_union) if len(pred_union) != 0 else 0)
#set_feature('successor_cosine', cosine_distance(data[source], data[target]))
#set_feature('predecessor_cosine', cosine_distance(source_pred, target_pred))
set_feature(
'shortest_path',
nx.shortest_path_length(digraph, source, target)
if digraph.has_edge(source, target) else 0)
pref_attch = nx.preferential_attachment(graph, [(source, target)])
for u, v, p in pref_attch:
set_feature('preference_attachment',
p) # if graph.has_edge(source, target) else 0)
aa_index = nx.adamic_adar_index(graph, [(source, target)])
for u, v, p in aa_index:
set_feature('adamic_adar_index',
p) # if graph.has_edge(source, target) else 0)
jcd_coe = nx.jaccard_coefficient(graph, [(source, target)])
for u, v, p in jcd_coe:
set_feature('jaccard_coefficient',
p) # if graph.has_edge(source, target) else 0)
reallo_index = nx.resource_allocation_index(
graph, [(source, target)])
for u, v, p in reallo_index:
set_feature('resource_allocation_index',
p) # if graph.has_edge(source, target) else 0)
set_feature('cluster_source', nx.clustering(graph, source))
set_feature('cluster_target', nx.clustering(graph, target))
set_feature('source_pagerank', pagerank[source])
set_feature('target_pagerank', pagerank[target])
set_feature('source_authorities', auth[source])
set_feature('target_authorities', auth[target])
set_feature('source_hubs', hub[source])
set_feature('target_hubs', hub[target])
#set_feature('source_core_num', core[source])
#set_feature('target_core_num', core[target])
except:
pass
return features
print("Number of edges deleted : %d" % edge_subset_G_size)
print("Number of edges remaining : %d" % (t - edge_subset_G_size))
#6 Create a train set of 80 percent from G_test
edges_to_remove_from_G_test = 0.201
removed_edges_from_G_test = random.sample(
G_test.edges(),
int(edges_to_remove_from_G_test * G_test.number_of_edges()))
G_train = G_test.copy()
G_train.remove_edges_from(removed_edges_from_G_test)
edge_subset_G_test_size = len(list(removed_edges_from_G_test))
print("Number of edges deleted : %d" % edge_subset_G_test_size)
print("Number of edges remaining : %d" %
(t - edge_subset_G_size - edge_subset_G_test_size))
#6 Transform G_train and G_test to undirected
G_train = G_train.to_undirected()
G_test = G_test.to_undirected()
#7 Calculate AA AUC
pred_aa_train = list(nx.adamic_adar_index(G_train))
pred_aa_test = list(nx.adamic_adar_index(G_test))
score_aa, label_aa = zip(*[(s, (u, v) in removed_edges_from_G)
for (u, v, s) in pred_aa_test])
auc_aa = roc_auc_score(label_aa, score_aa)
#8 Print AUC and prediciton calculation time
t2 = datetime.now()
delta = t2 - t1
print(auc_aa, delta.seconds)
def make_adamic_adar_index_predG(dirG, testG):
undG = dirG.to_undirected()
undir_AAs = nx.Graph()
undir_AAs.add_weighted_edges_from(nx.adamic_adar_index(undG, testG.edges_iter()))
return make_predG_from_jacc(undir_AAs, dirG, testG)
random.shuffle(mytrain)
del test[0]
g = nx.Graph()
connect = []
for i in range(20000):
t = train[i].split()
for j in range(len(t)):
g.add_edge(t[0], t[j])
for i in range(19999):
t1 = train[i].split()
t2 = train[i + 1].split()
if g.has_edge(t1[0], t2[0]):
connect.append([
len(list(nx.common_neighbors(g, t1[0], t2[0]))),
list(nx.adamic_adar_index(g, [(t1[0], t2[0])]))[0][2],
list(nx.preferential_attachment(g, [(t1[0], t2[0])]))[0][2],
list(nx.jaccard_coefficient(g, [(t1[0], t2[0])]))[0][2],
list(nx.resource_allocation_index(g, [(t1[0], t2[0])]))[0][2], 1
])
else:
connect.append([
len(list(nx.common_neighbors(g, t1[0], t2[0]))),
list(nx.adamic_adar_index(g, [(t1[0], t2[0])]))[0][2],
list(nx.preferential_attachment(g, [(t1[0], t2[0])]))[0][2],
list(nx.jaccard_coefficient(g, [(t1[0], t2[0])]))[0][2],
list(nx.resource_allocation_index(g, [(t1[0], t2[0])]))[0][2], 0
])
for i in range(300000):
t0, t1, t2 = mytrain[i].split()
def adamic_adar(graph):
output_file = open("data/imdb_b__adamic_adar", "w")
for (i, (u, v, score)) in enumerate(nx.adamic_adar_index(graph, graph.edges_iter())):
print i
output_file.write("\t".join(map(str, [u, v, score])) + "\n")
output_file.close()
#plt.plot(fpr, tpr, marker='.',label=leg)
plt.plot(recall,precision,marker='.',label=leg)
plt.legend()
return roc_score, ap_score,precision,recall
# ## 3. Adamic-Adar
# In[7]:
# Compute Adamic-Adar indexes from g_train
aa_matrix = np.zeros(adj.shape)
for u, v, p in nx.adamic_adar_index(g_train): # (u, v) = node indices, p = Adamic-Adar index
aa_matrix[u][v] = p
aa_matrix[v][u] = p # make sure it's symmetric
# Normalize array
aa_matrix = aa_matrix / aa_matrix.max()
# In[8]:
# Calculate ROC AUC and Average Precision
k=0
aa_roc, aa_ap,precision,recall= get_roc_score1(test_edges, test_edges_false, aa_matrix)
print ('Adamic-Adar Test ROC score: ', str(aa_roc))
# Drawing the Graph using matplotlib
nx.draw_networkx(UG, node_color=['red'], with_labels=True)
plt.show()
all_nodes = []
nodes = list(nx.nodes(UG))
# loop to append the pair of nodes to a list
for x in nodes:
for y in nodes:
if (x != y and not all_nodes.__contains__({x,y})):
all_nodes.append({x,y})
# Implementing Adamic-Adar
adamic_adar = nx.adamic_adar_index(UG, all_nodes) # the graph and the pair of nodes as parameters
# Print the values
print(" \nAdamic-Adar implementation \n")
max=0
for u, v, p in adamic_adar:
if(p>max):
max = p
sim_u = u
sim_v = v
print ('{}, {} -> {:.5f}'.format(u,v,p))
print (f'\nThe most similar according Adamic-Adar: {sim_u}, {sim_v} -> {max:.5f}')
# According to this results the highest Adamic-Adar similarity is between u2 and u3
linklist.append([int(data[0]), int(data[1])])
nodepair_set[random.randint(0, n_folds - 1)].append([int(data[0]), int(data[1])])
# new_line = data[0] + ' ' + data[1] + ' 1\n'
# f_w.write(new_line)
train_list = []
for templist in nodepair_set[0:8]:
train_list = train_list + templist
test_list = nodepair_set[9]
nodelist = create_vertex(linklist)
train_adj = create_adjmatrix(train_list, nodelist)
test_adj = create_adjmatrix(test_list, nodelist)
# print(train_adj)
sim_cn = np.dot(train_adj, train_adj)
sim_bifan = np.dot(np.dot(train_adj, train_adj.T), train_adj)
sim_AA = nx.adamic_adar_index(train_adj)
sim_RA = AA(train_adj)
sim_IP = IP(train_adj, 0.8)
# sim_jaccard = Jaccard(train_adj)
# cn_score_1 = AUC.Calculation_AUC(train_adj, test_adj, sim_cn, len(nodelist))
# cn_score_2 = evaluationMetric.cal_AUC(train_adj, test_adj, sim_cn, 10000)
cn_score_3 = metric.auc_score(sim_cn, test_adj, train_adj,'cc')
AA_score_3 = metric.auc_score(sim_AA, test_adj, train_adj, 'cc')
RA_score_3 = metric.auc_score(sim_RA, test_adj, train_adj, 'cc')
bifan_score_3 = metric.auc_score(sim_bifan, test_adj, train_adj,'cc')
IP_score = metric.auc_score(sim_IP, test_adj, train_adj, 'cc')
print(cn_score_3)
print(AA_score_3)
rr = []
# for i in corenodes:
# if i not in dataset:
# print(i)
# for test
# for i in corenodes:
# preds = nx.adamic_adar_index(G,nonedges(G,i))
# tenlargest = heapq.nlargest(100, preds, key = lambda x: x[2])
# for j in tenlargest:
# rr.append(j)
# print(len(rr)) #==21550
# result= heapq.nlargest(4000, rr, key = lambda x: x[2])
#for val
for i in val:
preds = nx.adamic_adar_index(G, nonedges(G, i))
tenlargest = heapq.nlargest(1000, preds, key=lambda x: x[2])
for j in tenlargest:
rr.append(j)
print(len(rr)) #==21550
result = heapq.nlargest(2000, rr, key=lambda x: x[2])
endtime = datetime.datetime.now()
print("time", (endtime - starttime).seconds)
count = 0
for i in result:
if i[2] == 0:
count += 1
print(count)
print(len(result))
N = G.number_of_nodes()
nodelist = list(G.nodes())
print(nx.info(G))
print(nx.number_of_nodes(G))
print(nx.number_of_edges(G))
print(nx.is_directed(G))
def nonedges(G, u): #a generator with (u,v) for every non neighbor v
for v in nx.non_neighbors(G, u):
yield (u, v)
for u in G.nodes():
adar = nx.adamic_adar_index(G, nonedges(G, u))
for v in nx.non_neighbors(G, u):
com = nx.common_neighbors(G, u, v)
jac = nx.jaccard_coefficient(G, nonedges(G, u))
res = nx.resource_allocation_index(G, nonedges(G, u))
pre = nx.preferential_attachment(G, nonedges(G, u))
allm = {m: {} for m in methods}
toponame = "datasetFourSquare" + ".csv"
with open(toponame, "w", newline="",
encoding="utf-8") as f: # binary mode for windows \r\n prob
writer = csv.writer(f, delimiter=',')
writer.writerow(['node1', 'node2'] + methods + ['label'])
for p in total_edges:
n1, n2 = p
# with open("adj_matrix.csv", "wb") as f:
# writer = csv.writer(f)
# writer.writerows(AJ_matrix)
AJ_matrix = nx.adjacency_matrix(G)
#print AJ_matrix(0,0)
print AJ_matrix.shape
# for ii in range(0, 1):
# tmp_missing_links = []
# for jj in range(0, len(node_sets)):
# if AJ_matrix[ii,jj] == 0:
# tmp_missing_links.append((int(final_nodes[ii]),int(final_nodes[jj])))
# print (int(final_nodes[ii]),int(final_nodes[jj]))
preds = nx.adamic_adar_index(G, ebunch=None)
print max(need_to_check)
#initial_node = 1
missing_links = []
old_u = 1
tmp_missing_links_score = []
tmp_missing_links_u = []
tmp_missing_links_v = []
for u, v, score in preds:
if int(u) in need_to_check[0:1000]:
#print u, old_u
if int(u) == old_u:
#print executed
tmp_missing_links_score.append(score)
def set_edge_weight(self, edge_weight_method='weight'):
if edge_weight_method == 'weight':
return
# Centrality based methods
elif edge_weight_method == 'edge_betweenness_centrality':
print("comptuing edge_betweenness_centrality..")
C = nx.edge_betweenness_centrality(self.G, weight='weight')
print("done!")
elif edge_weight_method == 'edge_betweenness_centrality_subset':
print("comptuing edge_betweenness_centrality_subset..")
C = nx.edge_current_flow_betweenness_centrality(self.G,
weight='weight')
print('done')
elif edge_weight_method == 'edge_current_flow_betweenness_centrality_subset':
print(
"comptuing edge_current_flow_betweenness_centrality_subset..")
C = nx.edge_current_flow_betweenness_centrality_subset(
self.G, weight='weight')
print('done')
elif edge_weight_method == 'edge_load_centrality':
print("comptuing edge_load_centrality..")
C = nx.edge_load_centrality(self.G)
print('done!')
# Link Prediction based methods
elif edge_weight_method == 'adamic_adar_index':
print("comptuing adamic_adar_index ..")
preds = nx.adamic_adar_index(self.G, self.G.edges())
C = {}
for u, v, p in preds:
C[(u, v)] = p
elif edge_weight_method == 'ra_index_soundarajan_hopcroft':
print("comptuing ra_index_soundarajan_hopcroft ..")
preds = nx.ra_index_soundarajan_hopcroft(self.G, self.G.edges())
C = {}
for u, v, p in preds:
C[(u, v)] = p
elif edge_weight_method == 'preferential_attachment':
print("comptuing preferential_attachment ..")
preds = nx.preferential_attachment(self.G, self.G.edges())
C = {}
for u, v, p in preds:
C[(u, v)] = p
#elif edge_weight_method=='cn_soundarajan_hopcroft':
# print("comptuing cn_soundarajan_hopcroft ..")
# preds=nx.cn_soundarajan_hopcroft(self.G,self.G.edges())
# C={}
# for u, v, p in preds:
# C[(u,v)]=p
elif edge_weight_method == 'within_inter_cluster':
print("comptuing within_inter_cluster ..")
preds = nx.within_inter_cluster(self.G, self.G.edges())
C = {}
for u, v, p in preds:
C[(u, v)] = p
elif edge_weight_method == 'resource_allocation_index':
print("comptuing resource allocation index ..")
preds = nx.resource_allocation_index(self.G, self.G.edges())
C = {}
for u, v, p in preds:
C[(u, v)] = p
elif edge_weight_method == 'jaccard_coefficient':
print("comptuing jaccard_coefficient..")
preds = nx.jaccard_coefficient(self.G, self.G.edges())
C = {}
for u, v, p in preds:
C[(u, v)] = p
print('done!')
for u, v, d in self.G.edges(data=True):
if edge_weight_method == None:
d['weight'] = 1
else:
d['weight'] = C[(u, v)]
return 1
def get_edge_weight(self, i, j):
aa_index = nx.adamic_adar_index(self._G, [(i, j)])
return six.next(aa_index)[2]
nodes = nx.nodes(DG)
edges = nx.edges(DG)
non_edges = nx.non_edges(DG)
'''Compute HAA, HJC and HRA'''
HAA = []
HJC = []
HRA = []
SD = []
for e in test_pairs:
if not DG.has_node(e[0]):
DG.add_node(e[0])
UDG.add_node(e[0])
if not DG.has_node(e[1]):
DG.add_node(e[1])
UDG.add_node(e[1])
AA = nx.adamic_adar_index(UDG, [e])
JC = nx.jaccard_coefficient(UDG, [e])
RA = nx.resource_allocation_index(UDG, [e])
spec_diff = DG.in_degree(e[1]) - DG.in_degree(
e[0]) # specificity_difference
SD.append(spec_diff)
try:
for u, v, p in AA:
HAA.append(p)
except ZeroDivisionError:
HAA.append(0)
pass
try:
for u, v, p in JC:
HJC.append(p)
positive_predictions_proba_slpc_DegCent = []
positive_predictions_proba_slpc_EigenCent = []
positive_predictions_proba_slpc_ClosenessCent = []
positive_predictions_proba_slpc_BetweenCent = []
positive_predictions_proba_slpc_PageRank = []
lenedg = len(pedges)
cntr = 0
for edge in pedges:
cntr += 1
print("\r {}/{}".format(cntr, lenedg), end="")
positive_predictions_proba_jcc.append(
list(nx.jaccard_coefficient(G, [edge]))[0][2])
positive_predictions_proba_ra.append(
list(nx.resource_allocation_index(G, [edge]))[0][2])
positive_predictions_proba_aa.append(
list(nx.adamic_adar_index(G, [edge]))[0][2])
positive_predictions_proba_pa.append(
list(nx.preferential_attachment(G, [edge]))[0][2])
positive_predictions_proba_cnsh.append(
list(nx.cn_soundarajan_hopcroft(
G, [edge]))[0][2]) # needs community information
positive_predictions_proba_rash.append(
list(nx.ra_index_soundarajan_hopcroft(
G, [edge]))[0][2]) # needs community information
positive_predictions_proba_wic.append(
list(nx.within_inter_cluster(
G, [edge]))[0][2]) # needs community information
positive_predictions_proba_slp_DegCent.append(
list(SLP_prediction(G, [edge], centrality="DegCent"))[0][2])
positive_predictions_proba_slp_EigenCent.append(
list(SLP_prediction(G, [edge], centrality="EigenCent"))[0][2])
return itertools.combinations(iterable, 2)
G = nx.read_edgelist("./data/drugbank_interactions.tsv",
delimiter="\t",
nodetype=str)
partition = community.best_partition(G)
nx.set_node_attributes(G, name='community', values=partition)
ap = list(all_pairs(G.nodes()))
cn = cn.cnbors(G, ap)
rai = nx.resource_allocation_index(G, ap)
jc = nx.jaccard_coefficient(G, ap)
aai = nx.adamic_adar_index(G, ap)
pa = nx.preferential_attachment(G, ap)
ccn = nx.cn_soundarajan_hopcroft(G, ap)
cra = nx.ra_index_soundarajan_hopcroft(G, ap)
wic = nx.within_inter_cluster(G, ap, community='community')
u, v, s1, s2, s3, s4, s5, s6, s7, s8, has_edge = ([] for i in range(11))
for m1, m2, m3, m4, m5, m6, m7, m8 in zip(cn, rai, jc, aai, pa, ccn, cra, wic):
u.append(m1[0])
v.append(m1[1])
s1.append(m1[2])
s2.append(m2[2])
s3.append(m3[2])
s4.append(m4[2])
s5.append(m5[2])
s6.append(m6[2])
def save_to_file_similarities(E1, j):
node_list = list(E1.nodes)
temp_short_path, temp_cn, temp_jc, temp_a, temp_pa = {}, {}, {}, {}, {}
#calculation start
for node1 in node_list:
for node2 in node_list:
temp = nx.jaccard_coefficient(E1, [(node1, node2)])
for u, v, p in temp:
temp_jc[u, v] = p
temp = nx.adamic_adar_index(E1, [(node1, node2)])
try:
for u, v, p in temp:
temp_a[u, v] = p
except:
temp_a[u, v] = 0.0
temp = nx.preferential_attachment(E1, [(node1, node2)])
for u, v, p in temp:
temp_pa[u, v] = p
temp = sorted(nx.common_neighbors(E1, node1, node2))
temp_common_neighbors = []
temp = sorted(nx.common_neighbors(E1, node1, node2))
for common_neighbor in temp:
temp_common_neighbors.append(common_neighbor)
temp_cn[node1, node2] = (len(temp_common_neighbors))
#Ill use Exception Checking cause we have DiGraph and maybe there is no path between to nodes
#so an error would compile
try:
length = nx.shortest_path_length(E1, node1, node2)
temp_short_path[node1, node2] = length
except:
temp_short_path[node1, node2] = 0.0
#calculation ends
temp_cn = OrderedDict(sorted(temp_cn.items(), key=lambda kx: kx[1]))
temp_short_path = OrderedDict(
sorted(temp_short_path.items(), key=lambda kx: kx[1]))
temp_jc = OrderedDict(sorted(temp_jc.items(), key=lambda kx: kx[1]))
temp_a = OrderedDict(sorted(temp_a.items(), key=lambda kx: kx[1]))
temp_pa = OrderedDict(sorted(temp_pa.items(), key=lambda kx: kx[1]))
directory = 'Subgraphs/Similarities/' + 'T' + str(j - 1) + '-' + 'T' + str(
j + 1) + '_E_' + 'T' + str(j - 1) + '-' + 'T' + str(j)
if not os.path.exists(directory):
os.makedirs(directory)
temp_path = temp_path = directory + '/' + str(E1.graph['name'])
save_as_csv(temp_path + '_shortest_path', temp_short_path,
['Node', 'Length of Shortest Path'])
save_as_csv(temp_path + '_jaccard_coef', temp_jc,
['(Node1, Node2)', 'Jaccard Coefficient'])
save_as_csv(temp_path + '_pref_attach.txt', temp_pa,
['(Node1, Node2)', 'Preferential Attachment'])
save_as_csv(temp_path + '_adamic_index.txt', temp_a,
['(Node1, Node2)', 'Adamic Index'])
save_as_csv(temp_path + '_common_neigh.txt', temp_cn,
['(Node1, Node2)', '# of Common Neighbors'])
def aa(G, i, j):
return sorted(nx.adamic_adar_index(G, [(i, j)]))[0][2]
def get_adamic(filepath):
D, pr, pr_df = get_pagerank(filepath, save_file=False)
H = D.to_undirected()
adm = nx.adamic_adar_index(H, ebunch=pr_df["relation"])
# adm_output=[(item[0],item[1],item[2]) for item in adm ]
pass
parts_idx = np.where(parts == i)[0]
parts_graph = train_G.subgraph(parts_idx)
# find top-1% nodes
landmarks = []
for node, degree in sorted(parts_graph.degree().items(), key=lambda item: item[1], reverse=True):
landmarks.append(node)
if len(landmarks) > parts_graph.number_of_nodes() * 0.01:
landmarks_in_train_G.extend(landmarks)
break
# save partial graph with original label
mapping = {n: parts_graph.node[n]['ori_label'] for n in parts_graph.nodes()}
parts_graph = nx.relabel_nodes(parts_graph, mapping, copy=True)
for (n, data) in parts_graph.nodes(data=True):
data.pop('ori_label', None)
with open(Path('clustering', dataset, '{}.gpickle'.format(i)), 'wb') as f:
nx.write_gpickle(parts_graph, f)
combinations = list(itertools.combinations(landmarks_in_train_G, 2))
combinations = set(tuple(sorted(item)) for item in combinations)
landmark_graph = nx.Graph()
for u, v, p in nx.adamic_adar_index(train_G, combinations):
ori_u = train_G.node[u]['ori_label']
ori_v = train_G.node[v]['ori_label']
edge_w = 2./(1 + np.exp(-p/10)) - 1.
landmark_graph.add_edge(ori_u, ori_v, attr_dict={'weight': edge_w})
landmark_graph.add_edge(ori_v, ori_u, attr_dict={'weight': edge_w})
with open(Path('clustering', dataset, 'landmark.gpickle'), 'wb') as f:
nx.write_gpickle(landmark_graph, f)
def adamicAdar(edges: np.array, output_file: str):
# Initialize graph.
graph = nx.read_edgelist("out_graph.txt", nodetype=int, create_using=nx.Graph())
preds = nx.adamic_adar_index(graph, edges)
RecommendationPolicies.writeNpToFile(output_file, preds)
def get_features(L, flag):
X = [[] for i in range(len(L))]
#=====================Social features(user-to-user graph)======================
#g0.adamic adar score
if flag['g0'] is True:
print("get feature g0")
preds = nx.adamic_adar_index(G, L)
cnt = 0
for (u, v, p) in preds:
X[cnt].append(p)
cnt += 1
#g1.jaccard coefficient
if flag['g1'] is True:
print("get feature g1")
preds = nx.jaccard_coefficient(G, L)
cnt = 0
for (u, v, p) in preds:
X[cnt].append(p)
cnt += 1
#g2.resource_allocation
if flag['g2'] is True:
print("get feature g2")
preds = nx.resource_allocation_index(G, L)
cnt = 0
for (u, v, p) in preds:
X[cnt].append(p)
cnt += 1
#g3.preferentail_attachment
if flag['g3'] is True:
print("get feature g3")
preds = nx.preferential_attachment(G, L)
cnt = 0
for (u, v, p) in preds:
X[cnt].append(p)
cnt += 1
#g4.shortest path length
if flag['g4'] is True:
print("get feature g4")
cnt = 0
for (u, v) in L:
if G.has_edge(u, v):
G.remove_edge(u, v)
if nx.has_path(G, u, v):
X[cnt].append(
nx.shortest_path_length(G, source=u, target=v) / 50000)
else:
X[cnt].append(1)
G.add_edge(u, v)
else:
if nx.has_path(G, u, v):
X[cnt].append(
nx.shortest_path_length(G, source=u, target=v) / 50000)
else:
X[cnt].append(1)
cnt += 1
#g5.common neighbors
if flag['g5'] is True:
print("get feature g5")
cnt = 0
for (u, v) in L:
if G.has_edge(u, v):
G.remove_edge(u, v)
T = [w for w in nx.common_neighbors(G, u, v)]
G.add_edge(u, v)
else:
T = [w for w in nx.common_neighbors(G, u, v)]
X[cnt].append(len(T))
cnt += 1
#g6.Approximate katz for social graph
if flag['g6'] is True:
print("get feature g6")
cnt = 0
for (u, v) in L:
p = 0
if G.has_edge(u, v):
G.remove_edge(u, v)
for x in G.neighbors(u):
for y in G.neighbors(v):
if x == y or G.has_edge(x, y):
p += 1
G.add_edge(u, v)
else:
for x in G.neighbors(u):
for y in G.neighbors(v):
if x == y or G.has_edge(x, y):
p += 1
X[cnt].append(p)
cnt += 1
#=========================checkin features=========================================
#c0.follower number
if flag['c0'] is True:
print("get feature c0")
cnt = 0
for (u, v) in L:
X[cnt].append(U[u]['follow_cnt'] * U[v]['follow_cnt']) # fu*fv
cnt += 1
#c1.same time same location
if flag['c1'] is True:
print("get feature c1")
cnt = 0
for (u, v) in L:
p = calculate_CCC(G, u, v)
X[cnt].append(p)
cnt += 1
#c2.same time same distinct spot
if flag['c2'] is True:
print("get deature c2")
cnt = 0
for (u, v) in L:
p = 0
dis_same_spot = []
for k in C[u]:
if k[1] not in dis_same_spot and k in C[v]:
dis_same_spot.append(k[1])
p += 1
X[cnt].append(p)
cnt += 1
#c3.same distinct spot (not necessarily same time)
if flag['c3'] is True:
cnt = 0
print("get feature c3")
for (u, v) in L:
p = 0
dis_same_spot = []
for k in C[u]:
if k[1] not in dis_same_spot:
for m in C[v]:
if k[1] == m[1]:
dis_same_spot.append(k[1])
p += 1
break
X[cnt].append(p)
cnt += 1
#c4.min Entropy
if flag['c4'] is True:
print("get feature c4")
cnt = 0
for (u, v) in L:
p = 0
E_list = []
for k in C[u]:
if k in C[v]:
spot = k[1]
if spot in S and S[spot]['entropy'] > 0:
E_list.append(S[spot]['entropy'])
if len(E_list) > 0:
p = min(E_list)
X[cnt].append(p)
cnt += 1
#c5. distance of mean_LL
if flag['c5'] is True:
cnt = 0
print("get feature c5")
for (u, v) in L:
dist = np.sqrt((U[u]['mean_LL'][0] - U[v]['mean_LL'][0])**2 +
(U[u]['mean_LL'][1] - U[v]['mean_LL'][1])**2)
X[cnt].append(dist)
cnt += 1
#c6.weighted same location
if flag['c6'] is True:
print("get feature c6")
cnt = 0
for (u, v) in L:
p = 0
for k in C[u]:
if k in C[v]:
spot = k[1]
#if spot in S and S[spot]['entropy'] > 0:
#p += 1/S[spot]['entropy']
if spot in S:
dist = np.sqrt(
(S[spot]['LL'][0] - U[u]['mean_LL'][0])**2 +
(S[spot]['LL'][1] - U[u]['mean_LL'][1])**2)
p += dist
dist = np.sqrt(
(S[spot]['LL'][0] - U[v]['mean_LL'][0])**2 +
(S[spot]['LL'][1] - U[v]['mean_LL'][1])**2)
p += dist
X[cnt].append(p)
cnt += 1
#c7.PP
if flag['c7'] is True:
print("get feature c7")
cnt = 0
for (u, v) in L:
p = len(C[u]) * len(C[v])
X[cnt].append(p)
cnt += 1
#c8.Total Common Friend Closeness (TCFC)
if flag['c8'] is True:
print("get feature c8")
cnt = 0
for (u, v) in L:
p = 0
if G.has_edge(u, v):
G.remove_edge(u, v)
for w in nx.common_neighbors(G, u, v):
T1 = [x for x in nx.common_neighbors(G, u, w)]
T2 = [x for x in nx.common_neighbors(G, v, w)]
p += len(T1) * len(T2)
G.add_edge(u, v)
else:
for w in nx.common_neighbors(G, u, v):
T1 = [x for x in nx.common_neighbors(G, u, w)]
T2 = [x for x in nx.common_neighbors(G, v, w)]
p += len(T1) * len(T2)
X[cnt].append(p)
cnt += 1
#c9.Total Common friend Checkin Count (TCFCC)
if flag['c9'] is True:
print("get feature c9")
cnt = 0
for (u, v) in L:
p = 0
if G.has_edge(u, v):
G.remove_edge(u, v)
for w in nx.common_neighbors(G, u, v):
p += calculate_CCC(G, u, w) * calculate_CCC(G, v, w)
G.add_edge(u, v)
else:
for w in nx.common_neighbors(G, u, v):
p += calculate_CCC(G, u, w) * calculate_CCC(G, v, w)
X[cnt].append(p)
cnt += 1
#c10. Common Category Checkin Counts Product (CCCP)
if flag['c10'] is True:
print("get feature c10")
cnt = 0
for (u, v) in L:
p = 0
for cat in U[u]['cate']:
if cat in U[v]['cate']:
p += U[u]['cate'][cat] * U[v]['cate'][cat]
X[cnt].append(p)
cnt += 1
#c11. Common Category Checkin Counts Product Ratio(CCCPR)
if flag['c11'] is True:
print("get feature c11")
cnt = 0
for (u, v) in L:
p = 0
u_cate_total = sum(U[u]['cate'][cat]**2 for cat in U[u]['cate'])
v_cate_total = sum(U[v]['cate'][cat]**2 for cat in U[v]['cate'])
for cat in U[u]['cate']:
if cat in U[v]['cate']:
p += (U[u]['cate'][cat] * U[v]['cate'][cat] /
np.sqrt(u_cate_total * v_cate_total))
X[cnt].append(p)
cnt += 1
#c12.trip route length all
if flag['c12'] is True:
print("get feature c12")
cnt = 0
for (u, v) in L:
tripDayLen1 = list()
tripDayLen2 = list()
tripDay = "starting"
tripLen = 0.0
lastSpot = [0.0, 0.0]
for k in C[u]:
if not (lastSpot[0] == 0.0 and lastSpot[1] == 0.0):
if k[1] in S:
tripLen += np.sqrt((lastSpot[0] -
S[k[1]]['LL'][0])**2 +
(lastSpot[1] - S[k[1]]['LL'][1])**2)
lastSpot[0] = S[k[1]]['LL'][0]
lastSpot[1] = S[k[1]]['LL'][1]
else:
if k[1] in S:
lastSpot[0] = S[k[1]]['LL'][0]
lastSpot[1] = S[k[1]]['LL'][1]
tripDay = "starting"
tripLen2 = 0.0
lastSpot = [0.0, 0.0]
for k in C[v]:
if not (lastSpot[0] == 0.0 and lastSpot[1] == 0.0):
if k[1] in S:
tripLen2 += np.sqrt(
(lastSpot[0] - S[k[1]]['LL'][0])**2 +
(lastSpot[1] - S[k[1]]['LL'][1])**2)
lastSpot[0] = S[k[1]]['LL'][0]
lastSpot[1] = S[k[1]]['LL'][1]
else:
if k[1] in S:
lastSpot[0] = S[k[1]]['LL'][0]
lastSpot[1] = S[k[1]]['LL'][1]
X[cnt].append(tripLen + tripLen2)
cnt += 1
#=========================Heter Graph features=====================================
#h0.Approximate katz for bipartite graph
if flag['h0'] is True:
print("get feature h0")
cnt = 0
for (u, v) in L:
p = 0
for x in B.neighbors(u):
for y in B.neighbors(v):
if x == y or B.has_edge(x, y):
p += 1
X[cnt].append(p)
cnt += 1
#h1.Approximate katz on HB
if flag['h1'] is True:
print("get feature h1")
cnt = 0
for (u, v) in L:
p = 0
if HB.has_edge(u, v):
HB.remove_edge(u, v)
for x in HB.neighbors(u):
for y in HB.neighbors(v):
if x == y or HB.has_edge(x, y):
p += 1
HB.add_edge(u, v)
else:
for x in HB.neighbors(u):
for y in HB.neighbors(v):
if x == y or HB.has_edge(x, y):
p += 1
X[cnt].append(p)
cnt += 1
#h2.Approximate katz on H
if flag['h2'] is True:
print("get feature h2")
cnt = 0
for (u, v) in L:
p = 0
if H.has_edge(u, v):
H.remove_edge(u, v)
for x in H.neighbors(u):
for y in H.neighbors(v):
if x == y or H.has_edge(x, y):
p += 1
H.add_edge(u, v)
else:
for x in H.neighbors(u):
for y in H.neighbors(v):
if x == y or H.has_edge(x, y):
p += 1
X[cnt].append(p)
cnt += 1
#h3.shortest path length on B
if flag['h3'] is True:
print("get feature h3")
cnt = 0
for (u, v) in L:
if nx.has_path(B, u, v):
X[cnt].append(
nx.shortest_path_length(B, source=u, target=v) / 50000)
else:
X[cnt].append(1)
cnt += 1
#h4.clustering coefiicient on H
if flag['h4'] is True:
print("get feature h4")
cnt = 0
for (u, v) in L:
if H.has_edge(u, v):
H.remove_edge(u, v)
p = nx.clustering(H, u) * nx.clustering(H, v)
H.add_edge(u, v)
else:
p = nx.clustering(H, u) * nx.clustering(H, v)
X[cnt].append(p)
cnt += 1
#h5. number of (user's loc friends)'s loc friends
if flag['h5'] is True:
print("get feature h5")
cnt = 0
for (u, v) in L:
counter1 = 0
for neighbor in H.neighbors(u):
if not neighbor.isnumeric():
for neighbor2 in H.neighbors(neighbor):
if not neighbor.isnumeric():
counter1 += 1
counter2 = 0
for neighbor in H.neighbors(v):
if not neighbor.isnumeric():
for neighbor2 in H.neighbors(neighbor):
if not neighbor.isnumeric():
counter2 += 1
#print(str(counter1)+" "+str(counter2)+"\n")
X[cnt].append(counter1 * counter2)
cnt += 1
return X
jaccard = np.zeros(n)
adar = np.zeros(n)
preferential_attachment = np.zeros(n)
resource_allocation_index = np.zeros(n)
common_neighbors = np.zeros(n)
# computing features for training set
for i in tqdm(range(len(id1))):
if training.at[str(id1[i]) + "|" + str(id2[i]), "target"] == 1:
G.remove_edge(id1[i], id2[i])
pred = nx.jaccard_coefficient(G, [(id1[i], id2[i])])
pred = [(u, v, p) for (u, v, p) in pred]
jaccard[i] = pred[0][2]
pred = nx.adamic_adar_index(G, [(id1[i], id2[i])])
pred = [(u, v, p) for (u, v, p) in pred]
adar[i] = pred[0][2]
pred = nx.preferential_attachment(G, [(id1[i], id2[i])])
pred = [(u, v, p) for (u, v, p) in pred]
preferential_attachment[i] = pred[0][2]
pred = nx.resource_allocation_index(G, [(id1[i], id2[i])])
pred = [(u, v, p) for (u, v, p) in pred]
resource_allocation_index[i] = pred[0][2]
pred = nx.common_neighbors(G, id1[i], id2[i])
pred = len([u for u in pred])
common_neighbors[i] = pred
def adamic_adar(graph, output_file_name):
outputFile = open(output_file_name + "_adamic_adar", 'w')
for (u, v, score) in nx.adamic_adar_index(graph, graph.edges()):
line = outputFormat(u, v, score)
outputFile.write(line)
outputFile.close()
# Common Neighbors
CN = [(e[0], e[1], len(list(nx.common_neighbors(M, e[0], e[1]))))
for e in nx.non_edges(M)]
CN.sort(key=operator.itemgetter(2), reverse=True)
# Jaccard coef
jaccard = list(nx.jaccard_coefficient(M))
jaccard.sort(key=operator.itemgetter(2), reverse=True)
# Resource Allocation index
RA = list(nx.resource_allocation_index(M))
RA.sort(key=operator.itemgetter(2), reverse=True)
# Adamic-Adar index
AA = list(nx.adamic_adar_index(M))
AA.sort(key=operator.itemgetter(2), reverse=True)
# Preferential Attachement
PA = list(nx.preferential_attachment(M))
PA.sort(key=operator.itemgetter(2), reverse=True)
# Community Common Neighbors !!! requires graph to have node attribute: 'community' !!!
#CCN = list(nx.cn_soundarajan_hopcroft(M))
#CCN.sort(key=operator.itemgetter(2), reverse = True)
# Community Resource Allocation !!! requires graph to have node attribute: 'community' !!!
#CRA = list(nx.ra_index_soundarajan_hopcroft(M))
#CRA.sort(key=operator.itemgetter(2), reverse = True)
# ###################### Prediction on Future Edge Linkage ####################
nx.draw_networkx(graph_1)
# In[18]:
plt.rcParams["figure.figsize"] = (20, 15)
nx.draw_networkx(graph_2,
node_size=np.array(list(nx.pagerank(graph_2, .5).values())) *
10**5)
# ## Prédiction de liens
# Évaluez la similarité, au sens de l'indice Adamic/Adar, entre toutes les paires de noeuds non adjacents du graphe non orienté `graph_1`
# In[19]:
print('\n[+] Exercice 8')
print(top_k_triplets(nx.adamic_adar_index(graph_1), 5))
# Implémentez la fonction `generic_common_neighbors` utilisée par `generic_adamic_adar`.
# In[20]:
def generic_common_neighbors(g, u, v):
"""
Intersection of u's neighbors and v's neighbors
:g: networkx graph
:u, v: str, nodes
:return: list, of common neighbors
"""
common = set(g.neighbors(u)).intersection(g.neighbors(v))
return list(common)
def gen_topol_feats(A_orig, A, edge_s):
"""
This function generates the topological features for matrix A (A_tr or A_ho) over edge samples edge_s (edge_tr or edge_ho).
Input and Parameters:
-------
A: the training or holdout adjacency matrix that the topological features are going to be computed over
A_orig: the original adjacency matrix
edge_s: the sample set of training or holdout edges that the topological features are going to be computed over
Returns:
-------
df_feat: data frame of features
Examples:
-------
>>> gen_topol_feats(A_orig, A_tr, edge_tr)
>>> gen_topol_feats(A_orig, A_ho, edge_ho)
"""
_, edges = adj_to_nodes_edges(A)
nodes = [int(iii) for iii in range(A.shape[0])]
N = len(nodes)
if len(edges.shape) == 1:
edges = [(int(iii), int(jjj)) for iii, jjj in [edges]]
else:
edges = [(int(iii), int(jjj)) for iii, jjj in edges]
# define graph
G = nx.Graph()
G.add_nodes_from(nodes)
G.add_edges_from(edges)
# average degree (AD)
ave_deg_net = np.sum(A) / A.shape[0]
# variance of degree distribution (VD)
var_deg_net = np.sqrt(
np.sum(np.square(np.sum(A, axis=0) - ave_deg_net)) / (A.shape[0] - 1))
# average (local) clustering coefficient (ACC)
ave_clust_net = nx.average_clustering(G)
# samples chosen - features
edge_pairs_f_i = edge_s[:, 0]
edge_pairs_f_j = edge_s[:, 1]
# local number of triangles for i and j (LNT_i, LNT_j)
numtriang_nodes_obj = nx.triangles(G)
numtriang_nodes = []
for nn in range(len(nodes)):
numtriang_nodes.append(numtriang_nodes_obj[nn])
numtriang1_edges = []
numtriang2_edges = []
for ee in range(len(edge_s)):
numtriang1_edges.append(numtriang_nodes[edge_s[ee][0]])
numtriang2_edges.append(numtriang_nodes[edge_s[ee][1]])
# Page rank values for i and j (PR_i, PR_j)
page_rank_nodes_obj = nx.pagerank(G)
page_rank_nodes = []
for nn in range(len(nodes)):
page_rank_nodes.append(page_rank_nodes_obj[nn])
page_rank1_edges = []
page_rank2_edges = []
for ee in range(len(edge_s)):
page_rank1_edges.append(page_rank_nodes[edge_s[ee][0]])
page_rank2_edges.append(page_rank_nodes[edge_s[ee][1]])
# j-th entry of the personalized page rank of node i (PPR)
page_rank_pers_nodes = []
hot_vec = {}
for nn in range(len(nodes)):
hot_vec[nn] = 0
for nn in range(len(nodes)):
hot_vec_copy = hot_vec.copy()
hot_vec_copy[nn] = 1
page_rank_pers_nodes.append(
nx.pagerank(G, personalization=hot_vec_copy))
page_rank_pers_edges = []
for ee in range(len(edge_s)):
page_rank_pers_edges.append(
page_rank_pers_nodes[edge_s[ee][0]][edge_s[ee][1]])
# local clustering coefficients for i and j (LCC_i, LCC_j)
clust_nodes_obj = nx.clustering(G)
clust_nodes = []
for nn in range(len(nodes)):
clust_nodes.append(clust_nodes_obj[nn])
clust1_edges = []
clust2_edges = []
for ee in range(len(edge_s)):
clust1_edges.append(clust_nodes[edge_s[ee][0]])
clust2_edges.append(clust_nodes[edge_s[ee][1]])
# average neighbor degrees for i and j (AND_i, AND_j)
ave_neigh_deg_nodes_obj = nx.average_neighbor_degree(G)
ave_neigh_deg_nodes = []
for nn in range(len(nodes)):
ave_neigh_deg_nodes.append(ave_neigh_deg_nodes_obj[nn])
ave_neigh_deg1_edges = []
ave_neigh_deg2_edges = []
for ee in range(len(edge_s)):
ave_neigh_deg1_edges.append(ave_neigh_deg_nodes[edge_s[ee][0]])
ave_neigh_deg2_edges.append(ave_neigh_deg_nodes[edge_s[ee][1]])
# degree centralities for i and j (DC_i, DC_j)
deg_cent_nodes_obj = nx.degree_centrality(G)
deg_cent_nodes = []
for nn in range(len(nodes)):
deg_cent_nodes.append(deg_cent_nodes_obj[nn])
deg_cent1_edges = []
deg_cent2_edges = []
for ee in range(len(edge_s)):
deg_cent1_edges.append(deg_cent_nodes[edge_s[ee][0]])
deg_cent2_edges.append(deg_cent_nodes[edge_s[ee][1]])
# eigenvector centralities for i and j (EC_i, EC_j)
tr = 1
toler = 1e-6
while tr == 1:
try:
eig_cent_nodes_obj = nx.eigenvector_centrality(G, tol=toler)
tr = 0
except:
toler = toler * 1e1
eig_cent_nodes = []
for nn in range(len(nodes)):
eig_cent_nodes.append(eig_cent_nodes_obj[nn])
eig_cent1_edges = []
eig_cent2_edges = []
for ee in range(len(edge_s)):
eig_cent1_edges.append(eig_cent_nodes[edge_s[ee][0]])
eig_cent2_edges.append(eig_cent_nodes[edge_s[ee][1]])
# Katz centralities for i and j (KC_i, KC_j)
ktz_cent_nodes_obj = nx.katz_centrality_numpy(G)
ktz_cent_nodes = []
for nn in range(len(nodes)):
ktz_cent_nodes.append(ktz_cent_nodes_obj[nn])
ktz_cent1_edges = []
ktz_cent2_edges = []
for ee in range(len(edge_s)):
ktz_cent1_edges.append(ktz_cent_nodes[edge_s[ee][0]])
ktz_cent2_edges.append(ktz_cent_nodes[edge_s[ee][1]])
# Jaccard’s coefficient of neighbor sets of i, j (JC)
jacc_coeff_obj = nx.jaccard_coefficient(G, edge_s)
jacc_coeff_edges = []
for uu, vv, jj in jacc_coeff_obj:
jacc_coeff_edges.append([uu, vv, jj])
df_jacc_coeff = pd.DataFrame(jacc_coeff_edges,
columns=['i', 'j', 'jacc_coeff'])
df_jacc_coeff['ind'] = df_jacc_coeff.index
# resource allocation index of i, j (RA)
res_alloc_ind_obj = nx.resource_allocation_index(G, edge_s)
res_alloc_ind_edges = []
for uu, vv, jj in res_alloc_ind_obj:
res_alloc_ind_edges.append([uu, vv, jj])
df_res_alloc_ind = pd.DataFrame(res_alloc_ind_edges,
columns=['i', 'j', 'res_alloc_ind'])
df_res_alloc_ind['ind'] = df_res_alloc_ind.index
# Adamic/Adar index of i, j (AA)
adam_adar_obj = nx.adamic_adar_index(G, edge_s)
adam_adar_edges = []
for uu, vv, jj in adam_adar_obj:
adam_adar_edges.append([uu, vv, jj])
df_adam_adar = pd.DataFrame(adam_adar_edges,
columns=['i', 'j', 'adam_adar'])
df_adam_adar['ind'] = df_adam_adar.index
df_merge = pd.merge(df_jacc_coeff,
df_res_alloc_ind,
on=['ind', 'i', 'j'],
sort=False)
df_merge = pd.merge(df_merge,
df_adam_adar,
on=['ind', 'i', 'j'],
sort=False)
# preferential attachment (degree product) of i, j (PA)
pref_attach_obj = nx.preferential_attachment(G, edge_s)
pref_attach_edges = []
for uu, vv, jj in pref_attach_obj:
pref_attach_edges.append([uu, vv, jj])
df_pref_attach = pd.DataFrame(pref_attach_edges,
columns=['i', 'j', 'pref_attach'])
df_pref_attach['ind'] = df_pref_attach.index
# global features:
# similarity of connections in the graph with respect to the node degree
# degree assortativity (DA)
deg_ass_net = nx.degree_assortativity_coefficient(G)
# transitivity: fraction of all possible triangles present in G
# network transitivity (clustering coefficient) (NT)
transit_net = nx.transitivity(G)
# network diameter (ND)
try:
diam_net = nx.diameter(G)
except:
diam_net = np.inf
ave_deg_net = [ave_deg_net for ii in range(10000)]
var_deg_net = [var_deg_net for ii in range(10000)]
ave_clust_net = [ave_clust_net for ii in range(10000)]
deg_ass_net = [deg_ass_net for ii in range(10000)]
transit_net = [transit_net for ii in range(10000)]
diam_net = [diam_net for ii in range(10000)]
com_ne = []
for ee in range(len(edge_s)):
com_ne.append(
len(sorted(nx.common_neighbors(G, edge_s[ee][0], edge_s[ee][1]))))
# closeness centralities for i and j (CC_i, CC_j)
closn_cent_nodes_obj = nx.closeness_centrality(G)
closn_cent_nodes = []
for nn in range(len(nodes)):
closn_cent_nodes.append(closn_cent_nodes_obj[nn])
closn_cent1_edges = []
closn_cent2_edges = []
for ee in range(len(edge_s)):
closn_cent1_edges.append(closn_cent_nodes[edge_s[ee][0]])
closn_cent2_edges.append(closn_cent_nodes[edge_s[ee][1]])
# shortest path between i, j (SP)
short_Mat_aux = nx.shortest_path_length(G)
short_Mat = {}
for ss in range(N):
value = next(short_Mat_aux)
short_Mat[value[0]] = value[1]
short_path_edges = []
for ee in range(len(edge_s)):
if edge_s[ee][1] in short_Mat[edge_s[ee][0]].keys():
short_path_edges.append(short_Mat[edge_s[ee][0]][edge_s[ee][1]])
else:
short_path_edges.append(np.inf)
# load centralities for i and j (LC_i, LC_j)
load_cent_nodes_obj = nx.load_centrality(G, normalized=True)
load_cent_nodes = []
for nn in range(len(nodes)):
load_cent_nodes.append(load_cent_nodes_obj[nn])
load_cent1_edges = []
load_cent2_edges = []
for ee in range(len(edge_s)):
load_cent1_edges.append(load_cent_nodes[edge_s[ee][0]])
load_cent2_edges.append(load_cent_nodes[edge_s[ee][1]])
# shortest-path betweenness centralities for i and j (SPBC_i, SPBC_j)
betw_cent_nodes_obj = nx.betweenness_centrality(G, normalized=True)
betw_cent_nodes = []
for nn in range(len(nodes)):
betw_cent_nodes.append(betw_cent_nodes_obj[nn])
betw_cent1_edges = []
betw_cent2_edges = []
for ee in range(len(edge_s)):
betw_cent1_edges.append(betw_cent_nodes[edge_s[ee][0]])
betw_cent2_edges.append(betw_cent_nodes[edge_s[ee][1]])
neigh_ = {}
for nn in range(len(nodes)):
neigh_[nn] = np.where(A[nn, :])[0]
df_pref_attach = []
for ee in range(len(edge_s)):
df_pref_attach.append(
len(neigh_[edge_s[ee][0]]) * len(neigh_[edge_s[ee][1]]))
U, sig, V = np.linalg.svd(A, full_matrices=False)
S = np.diag(sig)
Atilda = np.dot(U, np.dot(S, V))
Atilda = np.array(Atilda)
f_mean = lambda x: np.mean(x) if len(x) > 0 else 0
# entry i, j in low rank approximation (LRA) via singular value decomposition (SVD)
svd_edges = []
# dot product of columns i and j in LRA via SVD for each pair of nodes i, j
svd_edges_dot = []
# average of entries i and j’s neighbors in low rank approximation
svd_edges_mean = []
for ee in range(len(edge_s)):
svd_edges.append(Atilda[edge_s[ee][0], edge_s[ee][1]])
svd_edges_dot.append(
np.inner(Atilda[edge_s[ee][0], :], Atilda[:, edge_s[ee][1]]))
svd_edges_mean.append(
f_mean(Atilda[edge_s[ee][0], neigh_[edge_s[ee][1]]]))
# Leicht-Holme-Newman index of neighbor sets of i, j (LHN)
f_LHN = lambda num, den: 0 if (num == 0 and den == 0) else float(num) / den
LHN_edges = [
f_LHN(num, den)
for num, den in zip(np.array(com_ne), np.array(df_pref_attach))
]
U, sig, V = np.linalg.svd(A)
S = linalg.diagsvd(sig, A.shape[0], A.shape[1])
S_trunc = S.copy()
S_trunc[S_trunc < sig[int(np.ceil(np.sqrt(A.shape[0])))]] = 0
Atilda = np.dot(np.dot(U, S_trunc), V)
Atilda = np.array(Atilda)
f_mean = lambda x: np.mean(x) if len(x) > 0 else 0
# an approximation of LRA (LRA-approx)
svd_edges_approx = []
# an approximation of dLRA (dLRA-approx)
svd_edges_dot_approx = []
# an approximation of mLRA (mLRA-approx)
svd_edges_mean_approx = []
for ee in range(len(edge_s)):
svd_edges_approx.append(Atilda[edge_s[ee][0], edge_s[ee][1]])
svd_edges_dot_approx.append(
np.inner(Atilda[edge_s[ee][0], :], Atilda[:, edge_s[ee][1]]))
svd_edges_mean_approx.append(
f_mean(Atilda[edge_s[ee][0], neigh_[edge_s[ee][1]]]))
# number of nodes (N)
num_nodes = A_orig.shape[0]
# number of observed edges (OE)
num_edges = int(np.sum(A) / 2)
# construct a dictionary of the features
d = {'i':edge_pairs_f_i, 'j':edge_pairs_f_j, 'com_ne':com_ne, 'ave_deg_net':ave_deg_net, \
'var_deg_net':var_deg_net, 'ave_clust_net':ave_clust_net, 'num_triangles_1':numtriang1_edges, 'num_triangles_2':numtriang2_edges, \
'page_rank_pers_edges':page_rank_pers_edges, 'pag_rank1':page_rank1_edges, 'pag_rank2':page_rank2_edges, 'clust_coeff1':clust1_edges, 'clust_coeff2':clust2_edges, 'ave_neigh_deg1':ave_neigh_deg1_edges, 'ave_neigh_deg2':ave_neigh_deg2_edges,\
'eig_cent1':eig_cent1_edges, 'eig_cent2':eig_cent2_edges, 'deg_cent1':deg_cent1_edges, 'deg_cent2':deg_cent2_edges, 'clos_cent1':closn_cent1_edges, 'clos_cent2':closn_cent2_edges, 'betw_cent1':betw_cent1_edges, 'betw_cent2':betw_cent2_edges, \
'load_cent1':load_cent1_edges, 'load_cent2':load_cent2_edges, 'ktz_cent1':ktz_cent1_edges, 'ktz_cent2':ktz_cent2_edges, 'pref_attach':df_pref_attach, 'LHN':LHN_edges, 'svd_edges':svd_edges,'svd_edges_dot':svd_edges_dot,'svd_edges_mean':svd_edges_mean,\
'svd_edges_approx':svd_edges_approx,'svd_edges_dot_approx':svd_edges_dot_approx,'svd_edges_mean_approx':svd_edges_mean_approx, 'short_path':short_path_edges, 'deg_assort':deg_ass_net, 'transit_net':transit_net, 'diam_net':diam_net, \
'num_nodes':num_nodes, 'num_edges':num_edges}
# construct a dataframe of the features
df_feat = pd.DataFrame(data=d)
df_feat['ind'] = df_feat.index
df_feat = pd.merge(df_feat, df_merge, on=['ind', 'i', 'j'], sort=False)
return df_feat
def sort_edges_by_adamic_adar_index(graph, edges):
edges_sorted = sorted(list(nx.adamic_adar_index(graph, edges)), key=lambda l: l[
2], reverse=True, cmp=compare_with_ties)
return [(row[0], row[1]) for row in edges_sorted], [row[2] for row in edges_sorted]