def cluster_transi_UD():
f = FaceData()
G = f.create_graph() # 获取原始数据
count = 3
for i in xrange(100, 12000, 1000):
avg_cluster = []
avg_trans = []
avg_assort = []
if count > 0:
for j in xrange(5):
mh = child_graphs.impore_03(G, None, i, "unique")
avg_cluster.append(round(nx.average_clustering(mh), 4))
avg_trans.append(round(nx.transitivity(mh), 4))
avg_assort.append(
round(nx.degree_assortativity_coefficient(mh), 4))
a = sum(avg_cluster) / len(avg_cluster)
b = sum(avg_trans) / len(avg_trans)
c = sum(avg_assort) / len(avg_assort)
count -= 1
print float(i) / 1000, a, b, c
else:
mh = child_graphs.impore_03(G, None, i, "unique")
a = round(nx.average_clustering(mh), 4)
b = round(nx.transitivity(mh), 4)
c = round(nx.degree_assortativity_coefficient(mh), 4)
print float(i) / 1000, a, b, c
def cluster_a_RW():
f = FaceData()
G = f.create_graph() # 获取原始数据
count = 3
for i in xrange(100, 12000, 1000):
avg_cluster = []
avg_train = []
avg_assort = []
if count > 0:
for j in xrange(5):
rw = child_graphs.random_walk(G, None, i, "unique")
avg_cluster.append(round(nx.average_clustering(rw), 4))
avg_train.append(round(nx.transitivity(rw), 4))
avg_assort.append(
round(nx.degree_assortativity_coefficient(rw), 4))
a = sum(avg_cluster) / len(avg_cluster)
b = sum(avg_train) / len(avg_train)
c = sum(avg_assort) / len(avg_assort)
count -= 1
print float(i) / 1000, a, b, c
else:
rw = child_graphs.random_walk(G, None, i, "unique")
a = round(nx.average_clustering(rw), 4)
b = round(nx.transitivity(rw), 4)
c = round(nx.degree_assortativity_coefficient(rw), 4)
print float(i) / 1000, a, b, c
def sub_degree():
f = FaceData()
G = f.create_graph() # 获取原始数据
BFS = common.BFS(G, 10000)
_plt = degree.degree(G, plt, "Original", "k-") # 绘制原始网络的度分布
_plt = degree.ego_degree(G, BFS, _plt, "BFS", "rh-") # 随机爬虫子网的度分布
_plt.legend(loc="upper right") # 加入图例
_plt.show()
def ori_avg_degree():
f = FaceData()
G = f.create_graph() # 获取原始数据
rw = common.random_walk(G, None, 16000, "unique")
mhrw = common.metropolis_hastings_random_walk(G, None, 16000, "unique")
bfs = common.BFS(G, None, 16000, "unique")
print "原始网络的平均度为: ", degree.avg_degree(G)
print "RW抽样得到的平均度为: ", degree.avg_degree(G, rw)
print "MHRW抽样得到的平均度为: ", degree.avg_degree(G, mhrw)
print "BFS抽样得到的平均度为: ", degree.avg_degree(G, bfs)
def create_graph():
t = FaceData()
G = t.create_graph() # 获取原始数据
f = open("../compare/avg_degree/rw.txt", "w")
try:
rw = common.random_walk(G, None, 8000, "unique")
avg = float(sum([G.degree(i) for i in rw])) / len(rw)
f.write("RW_avgDegree ")
f.write(str(round(avg, 2)))
finally:
f.close()
def duplicate_epinions_Mhrw():
t = FaceData()
G = t.create_graph() # 获取原始数据
f = open("../compare/avg_degree/dup_Epinion_Mhrw.txt", "w")
try:
for i in xrange(100, 10000, 100):
mhrw = common.metropolis_hastings_random_walk(G, None, i, "unique")
f.write(str(i) + " " + str(mhrw) + "\n")
finally:
f.close()
def duplicate_twitter_UD():
t = FaceData()
G = t.create_graph() # 获取原始数据
f = open("../compare/avg_degree/dup_Twitter_UD.txt", "w")
try:
for i in xrange(100, 10000, 100):
mhrw = improve_MH.impore_02(G, None, i, "unique")
f.write(str(i) + " " + str(mhrw) + "\n")
finally:
f.close()
def sub_graph_assort():
f = FaceData()
G = f.create_graph() # 获取原始数据
rw = child_graphs.random_walk(G, None, 1000, "unique")
mhrw = child_graphs.metropolis_hastings_random_walk(
G, None, 1000, "unique")
print "原始网络的匹配系数: ", avg_cluster(G)
print "RW 子网的匹配系数: ", avg_cluster(G, rw)
print "MHRW 子网的匹配系数: ", avg_cluster(G, mhrw)
def sub_degree():
f = FaceData()
G = f.create_graph() # 获取原始数据
rw = child_graphs.random_walk(G, None, 10000, "unique")
mhrw = child_graphs.metropolis_hastings_random_walk(
G, None, 10000, "unique")
_plt = degree.degree(G, plt, "UNI", "b--") # 绘制原始网络的度分布
_plt = degree.degree(rw, _plt, "RW", "k") # 随机爬虫子网的度分布
_plt = degree.degree(mhrw, _plt, "MHRW", "r-.")
_plt.legend(loc="upper right") # 加入图例
_plt.show()
def transi_RW():
f = FaceData()
G = f.create_graph() # 获取原始数据
count = 3
for i in xrange(100, 12000, 1000):
avg_cluster = []
if count > 0:
for j in xrange(5):
rw = child_graphs.random_walk(G, None, i, "unique")
avg_cluster.append(round(nx.transitivity(rw), 4))
a = sum(avg_cluster) / len(avg_cluster)
count -= 1
print float(i) / 1000, a
else:
rw = child_graphs.random_walk(G, None, i, "unique")
a = round(nx.transitivity(rw), 4)
print float(i) / 1000, a
def compare_press():
f = FaceData()
G = f.create_graph() # 获取原始数据
rw = common.random_walk(G, None, 10000, "unique")
mhrw = common.metropolis_hastings_random_walk(G, None, 10000, "unique")
# print round(1/impore - 1, 4)
# 保存数据
f = open("../compare/alpha_dup/com_dup.txt", "w")
try:
f.write(str("rw") + " " + str(rw) + "\n")
f.write(str("mhrw") + " " + str(mhrw) + "\n")
finally:
f.close()
def multi_sampling():
f = FaceData()
G = f.create_graph() # 获取原始数据
_degree = dict() # 分析重组数据
chushu = 0
for i in xrange(38):
BFS = common.BFS(G, 10000)
for i in BFS:
_degree[G.degree(i)] = _degree.get(G.degree(i), 0) + 1
chushu += len(BFS)
print "step ", i
# 处理数据,取平均的度分布
x = sorted(_degree.iterkeys()) #生成x轴序列，从1到最大度
num = chushu
y = []
for i in x:
y.append(float(_degree[i]) / num)
# 保存数据
f = open("../compare/degree_plot/Epin_BFS.txt", "w")
try:
for i in x:
f.write(str(i) + " " + str(float(_degree[i]) / num) + "\n")
finally:
f.close()
def ego_degree():
f = FaceData()
G = f.create_graph() # 获取原始数据
# rw = common.random_walk(G,None,10000,"unique")
mhrw = common.metropolis_hastings_random_walk(G, None, 10000, "unique")
# bfs = common.BFS(G,None,8000,"unique")
impore = improve_MH.impore_03(G, None, 10000, "unique")
_plt = degree.degree(G, plt, "Original", "k-") # 绘制原始网络的度分布
# _plt = degree.ego_degree(G,bfs,_plt,"ego-bfs","c*")
# _plt = degree.ego_degree(G,rw,_plt,"ego-RW","k-")
# _plt = degree.ego_degree(G,mhrw,_plt,"MHRW","b-.")
_plt = degree.ego_degree(G, mhrw, _plt, "MHRW", "bv-")
# _plt = degree.ego_degree(G,impore,_plt,"UD","r:")
_plt = degree.ego_degree(G, impore, _plt, "UD", "rh-")
_plt.ylabel(r'$P(k_\upsilon=k) $')
_plt.xlabel(u'node degree k')
# _plt.xlim(0,100)
_plt.legend(loc="upper right") # 加入图例
_plt.show()
def alpha_press():
f = FaceData()
G = f.create_graph() # 获取原始数据
impore = improve_MH.impore_03(G, None, 10000, "unique")
# print round(1/impore - 1, 4)
# 保存数据
f = open("../compare/alpha_dup/a_dup.txt", "w")
try:
for i in xrange(5, 100, 5):
i = float(i) / 100
print i
dup = 0
for j in xrange(0, 10):
dup += improve_MH.impore_03(G, None, 10000, "unique", alpa=i)
dup /= 10
f.write(str(i) + " " + str(dup) + "\n")
print dup
finally:
f.close()
def __init__(self, x, limit):
f = FaceData()
self.G = f.create_graph() # 获取原始数据
self.plt = plt
self.x = x
self.limit = limit
def multi_sampling():
f = FaceData()
G = f.create_graph() # 获取原始数据
_degree = dict() # 分析重组数据
chushu = 0
for i in xrange(38):
mhrw = common.metropolis_hastings_random_walk(G, None, 10000, "unique")
# impore = improve_MH.impore_02(G, None, 10000, "unique")
for i in mhrw:
_degree[G.degree(i)] = _degree.get(G.degree(i), 0) + 1
chushu += len(mhrw)
print "step "
# 处理数据,取平均的度分布
x = sorted(_degree.iterkeys()) #生成x轴序列，从1到最大度
num = chushu
y = []
for i in x:
y.append(float(_degree[i]) / num)
# 保存数据
f = open("../compare/degree_plot/dup_Twitter_Mh.txt", "w")
try:
for i in x:
f.write(str(i) + " " + str(float(_degree[i]) / num) + "\n")
finally:
f.close()
##################
_degree = dict() # 分析重组数据
chushu = 0
for i in xrange(38):
# mhrw = common.metropolis_hastings_random_walk(G,None,10000,"unique")
impore = improve_MH.impore_02(G, None, 10000, "unique")
for i in impore:
_degree[G.degree(i)] = _degree.get(G.degree(i), 0) + 1
chushu += len(impore)
print "step "
# 处理数据,取平均的度分布
x = sorted(_degree.iterkeys()) #生成x轴序列，从1到最大度
num = chushu
y = []
for i in x:
y.append(float(_degree[i]) / num)
# 保存数据
f = open("../compare/degree_plot/dup_Twitter_Ud.txt", "w")
try:
for i in x:
f.write(str(i) + " " + str(float(_degree[i]) / num) + "\n")
finally:
f.close()
f = FaceData()
G = f.create_graph() # 获取原始数据
count = 3
for i in xrange(100, 12000, 1000):
avg_cluster = []
avg_trans = []
avg_assort = []
if count > 0:
for j in xrange(5):
mh = child_graphs.BFS(G, i)
avg_cluster.append(round(nx.average_clustering(mh), 4))
avg_trans.append(round(nx.transitivity(mh), 4))
avg_assort.append(
round(nx.degree_assortativity_coefficient(mh), 4))
a = sum(avg_cluster) / len(avg_cluster)
b = sum(avg_trans) / len(avg_trans)
c = sum(avg_assort) / len(avg_assort)
count -= 1
print float(i) / 1000, a, b, c
else:
mh = child_graphs.BFS(G, i)
a = round(nx.average_clustering(mh), 4)
b = round(nx.transitivity(mh), 4)
c = round(nx.degree_assortativity_coefficient(mh), 4)
print float(i) / 1000, a, b, c
# cluster_transi_BFS()
f = FaceData()
G = f.create_graph()
print round(nx.degree_assortativity_coefficient(G), 4)
def bfs_text():
f = FaceData()
G = f.create_graph() # 获取原始数据
bfs = common.BFS(G, None, 5000, "total")
print len(bfs)
