-
Notifications
You must be signed in to change notification settings - Fork 0
/
robustnessnx_less.py
219 lines (187 loc) · 10.3 KB
/
robustnessnx_less.py
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
# -*- coding: utf-8 -*-
"""
Created on Sun Jul 19 01:35:18 2020
@author: Irene López-Rodríguez
Input variables:
type_method = [0 : random, 1 : degree]
projection = [0 : ICD, 1 : ATC]
"""
import sys
import networkx as nx
from networkx.algorithms import bipartite
import collections
import pandas as pd
import random
import numpy as np
#import matplotlib
#import matplotlib.pyplot as plt
def nodestoremove(p, q, nodes_0_c, nodes_1_c, df_atc, df_icd, type_method):
if type_method == 0:
atcNodesToRemove = random.sample(nodes_1_c, int(round(p * len(nodes_1_c))))
icdNodesToRemove = random.sample(nodes_0_c, int(round(q * len(nodes_0_c))))
elif type_method == 1:
atcNodesToRemove = list(df_atc.head(int(round(p * len(nodes_1_c))))['node'])
icdNodesToRemove = list(df_icd.head(int(round(q * len(nodes_0_c))))['node'])
return icdNodesToRemove, atcNodesToRemove
if __name__ == '__main__':
# Remover al azar 0 o dirigido 1
# Proyección ICD 0 o ATC 1
type_method = int(sys.argv[1])
type_proj = int(sys.argv[2])
print("Reading file ...")
vdmdata = pd.read_csv('vdmdata_reduce.csv', encoding = 'utf-8-sig')
nodes_0 = []
nodes_1 = []
for m in vdmdata.iterrows():
nodes_0.append(m[1][0]) #ICD
nodes_1.append(m[1][1]) #ATC
nodes_0 = list(dict.fromkeys(nodes_0))
nodes_1 = list(dict.fromkeys(nodes_1))
print("Building a bipartite graph ...")
# Build a bipartite graph:
G = nx.Graph()
# Add nodes ATC - ICD
G.add_nodes_from(nodes_0, bipartite=0) # Add the node attribute “bipartite” disease
G.add_nodes_from(nodes_1, bipartite=1) # active substance
# Add edges without weight
for m in vdmdata.iterrows():
enfermedad = m[1][0];
#peso = m[1][3];
sustancia = m[1][1];
G.add_edge(enfermedad, sustancia)
print("Getting largest component ...")
components = sorted(nx.connected_components(G), key=len, reverse=True)
largest_component = components[0]
C = G.subgraph(largest_component)
degX,degY=bipartite.degrees(C,nodes_0)
degATC = dict(degX).values()
degCIE = dict(degY).values()
counterATC = collections.Counter(degATC)
counterCIE = collections.Counter(degCIE)
df_icd = pd.DataFrame(dict(degY).items(), columns=['node', 'degree'])
df_atc = pd.DataFrame(dict(degX).items(), columns=['node', 'degree'])
df_icd = df_icd.sort_values(by=['degree', 'node'], ascending=False)
df_atc = df_atc.sort_values(by=['degree', 'node'], ascending=False)
nodes_0_c = []
nodes_1_c = []
for n in C.nodes(data=True):
if n[1]['bipartite'] == 0:
nodes_0_c.append(n[0])
if n[1]['bipartite'] == 1:
nodes_1_c.append(n[0])
# unfrozen_graph = nx.Graph(C)
print("Defining variables ...")
# p_vector = np.arange(0,1,0.1)
# p_vector = [ round(x * 0.001, 3) for x in range(0, 1000)]
p_vector = [ round(x * 0.1, 1) for x in range(0, 10)]
# [ round(x * 0.01, 2) for x in range(0, 100)]
np_avg_degree = np.zeros((len(p_vector), len(p_vector)))
np_conn_nodes = np.zeros((len(p_vector), len(p_vector)))
np_unconn_nodes = np.zeros((len(p_vector), len(p_vector)))
np_conn_components = np.zeros((len(p_vector), len(p_vector)))
np_mean_size = np.zeros((len(p_vector), len(p_vector)))
# np_clustering = np.zeros((len(p_vector), len(p_vector)))
# type_method = 0
# type_proj = 1
if type_method == 0:
for i in range(1,11):
print("Runing iteration random ... "+str(i))
index_p = 0
index_q = 0
for p in p_vector:
for q in p_vector:
unfrozen_graph = nx.Graph(C)
icdNodesToRemove, atcNodesToRemove = nodestoremove(p, q, nodes_0_c, nodes_1_c, df_atc, df_icd, type_method)
unfrozen_graph.remove_nodes_from(icdNodesToRemove)
unfrozen_graph.remove_nodes_from(atcNodesToRemove)
if type_proj == 0:
nodes_lst = [x for x in nodes_0_c if x not in icdNodesToRemove]
elif type_proj == 1:
nodes_lst = [x for x in nodes_1_c if x not in atcNodesToRemove]
GP = bipartite.projected_graph(unfrozen_graph, nodes_lst)
components = sorted(nx.connected_components(GP), key=len, reverse=True)
nodes_connected = sum(list(map(lambda c: len(c), components)))
mean_size_components = nodes_connected / len(components)
nodes_unconnected = len(nodes_0_c) - nodes_connected
degrees = GP.degree()
sum_of_edges = sum(list(dict(degrees).values()))
avg_degree = sum_of_edges / GP.number_of_nodes()
# avg_clustering = nx.average_clustering(GP)
if i == 1:
np_avg_degree[index_p][index_q] = avg_degree
np_conn_nodes[index_p][index_q] = nodes_connected
np_unconn_nodes[index_p][index_q] = nodes_unconnected
np_conn_components[index_p][index_q] = len(components)
np_mean_size[index_p][index_q] = mean_size_components
# np_clustering[index_p][index_q] = avg_clustering
else:
np_avg_degree[index_p][index_q] = np_avg_degree[index_p][index_q] + avg_degree
np_conn_nodes[index_p][index_q] = np_conn_nodes[index_p][index_q] + nodes_connected
np_unconn_nodes[index_p][index_q] = np_unconn_nodes[index_p][index_q] + nodes_unconnected
np_conn_components[index_p][index_q] = np_conn_components[index_p][index_q] + len(components)
np_mean_size[index_p][index_q] = np_mean_size[index_p][index_q] + mean_size_components
# np_clustering[index_p][index_q] = np_clustering[index_p][index_q] + avg_clustering
index_q += 1
index_q = 0
index_p += 1
elif type_method == 1:
index_p = 0
index_q = 0
i = 1
print("Runing directed ...")
for p in p_vector:
for q in p_vector:
unfrozen_graph = nx.Graph(C)
icdNodesToRemove, atcNodesToRemove = nodestoremove(p, q, nodes_0_c, nodes_1_c, df_atc, df_icd, type_method)
unfrozen_graph.remove_nodes_from(icdNodesToRemove)
unfrozen_graph.remove_nodes_from(atcNodesToRemove)
if type_proj == 0:
nodes_lst = [x for x in nodes_0_c if x not in icdNodesToRemove]
elif type_proj == 1:
nodes_lst = [x for x in nodes_1_c if x not in atcNodesToRemove]
GP = bipartite.projected_graph(unfrozen_graph, nodes_lst)
components = sorted(nx.connected_components(GP), key=len, reverse=True)
nodes_connected = sum(list(map(lambda c: len(c), components)))
mean_size_components = nodes_connected / len(components)
nodes_unconnected = len(nodes_0_c) - nodes_connected
degrees = GP.degree()
sum_of_edges = sum(list(dict(degrees).values()))
avg_degree = sum_of_edges / GP.number_of_nodes()
# avg_clustering = nx.average_clustering(GP)
if i == 1:
np_avg_degree[index_p][index_q] = avg_degree
np_conn_nodes[index_p][index_q] = nodes_connected
np_unconn_nodes[index_p][index_q] = nodes_unconnected
np_conn_components[index_p][index_q] = len(components)
np_mean_size[index_p][index_q] = mean_size_components
# np_clustering[index_p][index_q] = avg_clustering
else:
np_avg_degree[index_p][index_q] = np_avg_degree[index_p][index_q] + avg_degree
np_conn_nodes[index_p][index_q] = np_conn_nodes[index_p][index_q] + nodes_connected
np_unconn_nodes[index_p][index_q] = np_unconn_nodes[index_p][index_q] + nodes_unconnected
np_conn_components[index_p][index_q] = np_conn_components[index_p][index_q] + len(components)
np_mean_size[index_p][index_q] = np_mean_size[index_p][index_q] + mean_size_components
# np_clustering[index_p][index_q] = np_clustering[index_p][index_q] + avg_clustering
index_q += 1
index_q = 0
index_p += 1
np_avg_degree = np.true_divide(np_avg_degree, i)
np_conn_nodes = np.true_divide(np_conn_nodes, i)
np_unconn_nodes = np.true_divide(np_unconn_nodes, i)
np_conn_components = np.true_divide(np_conn_components, i)
np_mean_size = np.true_divide(np_mean_size, i)
# np_clustering = np.true_divide(np_clustering, i)
print("Creating files ... ")
df = pd.DataFrame(np_avg_degree, index=p_vector, columns=p_vector)
df.to_csv('np_avg_degree_2_'+str(type_proj)+'_'+str(type_method)+'.csv', index=True, header=True, sep=',', encoding = 'utf-8-sig')
df = pd.DataFrame(np_conn_nodes, index=p_vector, columns=p_vector)
df.to_csv('np_conn_nodes_2_'+str(type_proj)+'_'+str(type_method)+'.csv', index=True, header=True, sep=',', encoding = 'utf-8-sig')
df = pd.DataFrame(np_unconn_nodes, index=p_vector, columns=p_vector)
df.to_csv('np_unconn_nodes_2_'+str(type_proj)+'_'+str(type_method)+'.csv', index=True, header=True, sep=',', encoding = 'utf-8-sig')
df = pd.DataFrame(np_conn_components, index=p_vector, columns=p_vector)
df.to_csv('np_conn_components_2_'+str(type_proj)+'_'+str(type_method)+'.csv', index=True, header=True, sep=',', encoding = 'utf-8-sig')
df = pd.DataFrame(np_mean_size, index=p_vector, columns=p_vector)
df.to_csv('np_mean_size_2_'+str(type_proj)+'_'+str(type_method)+'.csv', index=True, header=True, sep=',', encoding = 'utf-8-sig')
# df = pd.DataFrame(np_clustering, index=p_vector, columns=p_vector)
# df.to_csv('np_clustering_'+str(type_proj)+'_'+str(type_method)+'.csv', index=True, header=True, sep=',', encoding = 'utf-8-sig')
# np.true_divide(np_avg_degree, i)