-
Notifications
You must be signed in to change notification settings - Fork 2
/
project_parallel.py
252 lines (197 loc) · 7.5 KB
/
project_parallel.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
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Fri Nov 9 17:03:12 2018
@author: s2077981
"""
import numpy as np
import networkx as nx
from time import time
import random
from sklearn.metrics import f1_score, precision_recall_curve, precision_recall_fscore_support
import pandas as pd
from multiprocessing import Pool
import operator
import os
import sys
import matplotlib.pyplot as plt
plt.switch_backend('agg')
from tqdm import tqdm
from joblib import Parallel, delayed
import multiprocessing
#import numba as nb
###################################################
os.chdir("/home/dead/Documents/SNACS/Snacs_Final")
#os.chdir("/home/melidell024/Desktop/snacs/project/Incremental-PageRank/")
####################################################
from incremental import IncrementalPersonalizedPageRank2 as inc
from retrieval_metrics import mean_average_precision,plot_precision
"""
Gnutella Fasoula
Nodes 6301
Edges 20777
RoadsPA
Nodes: 1088092
Edges: 1541898
"""
# Fixing the dataset format
def preprocess(raw_data,head=0):
# Input: path of a raw dataset
# Output: data_path of the corrected data
# Export: Correct Dataset
os.chdir("datasets/")
if not os.path.isfile('corrected'+raw_data):
start = time()
pan = pd.read_csv(raw_data, sep="\t", header=head)
#pan = pan.drop(0, axis=0)
#pan.columns= ["Source", "Target"]
data_path = 'corrected'+raw_data
pan.to_csv(data_path, header=False, sep='\t', index = False )
os.chdir("../")
print("Time taken for Preprocess: " + str(time()-start))
print("Dataset preprocessed and exported >_")
return data_path
else:
data_path = 'corrected'+raw_data
print("Dataset already preprocessed >_\n")
os.chdir("../")
return data_path
#data_path = preprocess('p2p-Gnutella08.txt')
####################################################
#@nb.jit
def Import(datapath, discriptives=False, directed=True):
# Function for importing the dataset into networkx
# Print some statistics
if directed:
graph_type = nx.DiGraph()
else:
graph_type = nx.Graph()
start = time()
with open('datasets/'+datapath, 'rb') as inf:
data = nx.read_edgelist(inf,create_using=graph_type,\
delimiter='\t', encoding="utf-8")
print("Time taken for edgelist loading: %.2fsec" % (time()-start))
# Discriptives
if discriptives:
print('\n-------------------------')
print('DESCRIPTIVE_STATISTICS')
print('-------------------------')
print('Is Graph directed?: ' + str(nx.is_directed(data)))
print('Number of nodes: %i' % data.number_of_nodes())
print('Number of edges: %i' % data.number_of_edges())
print('Density: %.5f' % nx.density(data))
print('-------------------------\n')
return data
# Pagerank
#@nb.jit(parallel=True)
def PPR (data, node, maxiter=500, alpha=0.7):
#start = time()
truePPR = nx.pagerank(data, alpha=alpha, personalization={node: 1}, max_iter=maxiter)
#print("\nTime taken for PageRank computation: %.2fsec" % (time()-start))
return truePPR
#@nb.jit(parallel=True)
def Approximate(data, node, n_walks=1000):
#start = time()
increment = inc(graph=data, node=node, number_of_random_walks=n_walks)
increment.initial_random_walks()
hat_PPR = increment.compute_personalized_page_ranks()
#print('\nTime taken for Approximation: %.2fsec' % (time()-start))
return hat_PPR
#@nb.jit
def Evaluate_values(true,pred):
true = sorted(true.items(), key=operator.itemgetter(0), reverse=True)
pred = sorted(pred.items(), key=operator.itemgetter(0), reverse=True)
true_values = [v for _,v in true]
pred_values = [v for _,v in pred]
true = np.array(true_values)
pred = np.array(pred_values)
MSE = np.mean((true-pred)**2)
RMSE = np.sqrt(MSE)
MAE = np.mean(np.abs((true-pred)))
eucl = np.linalg.norm(true-pred)
eucl_norm = np.linalg.norm(true-pred)/np.linalg.norm(pred)
lala = np.max(true-pred)/np.max(true)
cor = np.corrcoef(true,pred)
return MAE, RMSE, eucl, eucl_norm, lala, cor[0,1]
#@nb.jit
def Evaluate_retrieval(true,pred, k):
true = sorted(true.items(), key=operator.itemgetter(1), reverse=True)
pred = sorted(pred.items(), key=operator.itemgetter(1), reverse=True)
true_nodes = [n for n,_ in true[0:k]]
pred_nodes = [n for n,_ in pred[0:k*10]]
#ff1 = f1_score(true_nodes, pred_nodes, average='micro')
true = set(true_nodes)
pred = set(pred_nodes[0:k])
#ranked retrieval
true_r = np.array(pred_nodes)
pred_r = np.array(true_nodes)
retrieval_array = np.isin(true_r,pred_r)
#precision = r_precision(retrieval_array)
#avg_precision = average_precision(retrieval_array)
#m_avg_precision = mean_average_precision(list(retrieval_array))
#print("precision:",precision)
# print("average precision:",avg_precision)
#print("out :",len(retrieval_array))
#print("mean average precision:",m_avg_precision)
#Acc = len(true.intersection(pred))/(len(true))
jac = len(true.intersection(pred))/len(true.union(pred))
return retrieval_array, jac # kai oti allo valoume edw:P
#@nb.jit(parallel=True)
def mean_statistics(results,k):
v =[]
r = []
r_jacc=[]
for i in range(20): v.append(results[i][0][0]); r.append(results[i][1][0]), r_jacc.append(results[i][2][0])
res = np.array(v)
#retrieval = np.array(r)
avg_v = np.mean(res, axis=0)
Map = mean_average_precision(r)#[0]
#poutses = mean_average_precision(r)[1]
poutses,outs = plot_precision(r,k)
avg_jacc = np.mean(r_jacc)
return avg_v, Map, poutses, avg_jacc,outs
def plotting(x):
plt.figure(figsize=(8,6))
plt.tick_params(size=5, labelsize=15)
plt.plot(x)
plt.xlabel("Recall", fontsize=15)
plt.ylabel("Precission", fontsize=15)
plt.title("Mean Interpolated Average Precission", fontsize=15)
plt.savefig(dataset+'/plot.png')
plt.show()
def lolen(node):
true = PPR(data, node=node)
hat = Approximate(data,node=node)
results_v.append(Evaluate_values(true,hat))
results_retrieval.append(Evaluate_retrieval(true,hat,k)[0])
results_jaccard.append(Evaluate_retrieval(true,hat,k)[1])
return results_v, results_retrieval, results_jaccard
if __name__=='__main__':
data = ['p2p-Gnutella08.txt']
k=100
runs = 20
for dataset in data:
print('')
print("Network-Dataset: || %s ||" % dataset)
print("-----------------------------------------------")
if not os.path.exists(dataset):
os.mkdir(dataset)
results_retrieval = []
results_jaccard = []
results_v = []
data_path = preprocess(dataset, head=None)
if dataset == 'cit-HepPh.txt' or dataset=='sx-superuser.txt': direct = True
else: direct = False
data = Import(data_path, discriptives=True, directed=direct)
node = random.sample(list(data.nodes()),20)
results = Parallel(n_jobs=4)(delayed(lolen)(i) for i in node)
avg_v, Map, p, avg_jacc,outs= mean_statistics(results,k)
plotting(p)
np.save(dataset+'/Map', Map)
np.save(dataset+'/avg_v', avg_v)
np.save(dataset+'/retrieval', np.array(results_retrieval))
np.save(dataset+'/statistics', np.array(results_v))
print("\n-------------------------")
print(">_ SUPPORT GNU/Linux >_")
print("-------------------------\n")
print('')