-
Notifications
You must be signed in to change notification settings - Fork 10
/
gen_survival_data.py
189 lines (140 loc) · 7.17 KB
/
gen_survival_data.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
import numpy as np
import json
from collections import defaultdict
from safdKit import sample_shuffle_uspv, order_grpah, acc_pair
from keras.preprocessing.sequence import pad_sequences
source_path = "../Twitter/Data/"
dest_path = "./ori_data/"
# users & labels
users, labels = np.load(source_path + "extract_data/users.npy"), np.load(source_path + "extract_data/labels.npy")
uncen_usr, cen_usr = (users[labels == 0]).tolist(), (users[labels == 1]).tolist()
#1. Observed Time
followers_count = json.load(open(source_path + "extract_data/followers_count.json", "r"))
usr_2_T = defaultdict(list)
for usr, followers in followers_count.items():
usr_2_T[usr] = len(followers)
#2. construct survival analysis data set with 'uncensored & censored' users above.
friends_count = json.load(open(source_path + "extract_data/friends_count.json", "r"))
listed_count = json.load(open(source_path + "extract_data/listed_count.json", "r"))
favourites_count = json.load(open(source_path + "extract_data/favourites_count.json", "r"))
statuses_count = json.load(open(source_path + "extract_data/statuses_count.json", "r"))
uncen_usr_seq, cen_usr_seq = defaultdict(list), defaultdict(list)
for usr in uncen_usr:
try:
uncen_usr_seq[usr]\
= [[map(int, followers_count[usr]), map(int,friends_count[usr]),
map(int, listed_count[usr]),map(int, favourites_count[usr]),
map(int,statuses_count[usr])], usr_2_T[usr],1]
except ValueError:
uncen_usr.remove(usr)
for usr in cen_usr:
try:
cen_usr_seq[usr]\
= [[map(int,followers_count[usr]), map(int,friends_count[usr]),
map(int,listed_count[usr]),map(int,favourites_count[usr]),
map(int,statuses_count[usr])], usr_2_T[usr],0]
except ValueError:
cen_usr.remove(usr)
for u, v in uncen_usr_seq.items():
uncen_usr_seq[u][0] = pad_sequences(v[0], maxlen=22, padding="post")
for u, v in cen_usr_seq.items():
cen_usr_seq[u][0] = pad_sequences(v[0], maxlen=22, padding="post")
UU_uncen = list()
UU_cen = list()
X_uncen = list()
X_cen = list()
T_uncen = list()
T_cen = list()
C_uncen = list()
C_cen = list()
for u, v in uncen_usr_seq.items():
UU_uncen.append(u)
X_uncen.append(np.array(v[0]).transpose())
T_uncen.append(v[1])
C_uncen.append(v[2])
for u, v in cen_usr_seq.items():
UU_cen.append(u)
X_cen.append(np.array(v[0]).transpose())
T_cen.append(v[1])
C_cen.append(v[2])
def ran_seed(N):
s = np.arange(N)
np.random.shuffle(s)
return s
UU_uncen, X_uncen, T_uncen, C_uncen = np.array(UU_uncen), np.array(X_uncen), np.array(T_uncen), np.array(C_uncen)
s_uncen = ran_seed(UU_uncen.shape[0])
UU_uncen, X_uncen, T_uncen, C_uncen = UU_uncen[s_uncen], X_uncen[s_uncen], T_uncen[s_uncen], C_uncen[s_uncen]
UU_uncen_sampling = list()
X_uncen_sampling = list()
T_uncen_sampling = list()
C_uncen_sampling = list()
for t in set(T_uncen):
UU_uncen_sampling.extend(UU_uncen[T_uncen == t])
X_uncen_sampling.extend(X_uncen[T_uncen == t])
T_uncen_sampling.extend(T_uncen[T_uncen == t])
C_uncen_sampling.extend(C_uncen[T_uncen == t])
UU_uncen_sampling, X_uncen_sampling, \
T_uncen_sampling, C_uncen_sampling = np.array(UU_uncen_sampling), np.array(X_uncen_sampling), \
np.array(T_uncen_sampling), np.array(C_uncen_sampling)
U_train, X_train, T_train, C_train = list(), list(), list(), list()
U_valid, X_valid, T_valid, C_valid = list(), list(), list(), list()
U_test, X_test, T_test, C_test = list(), list(), list(), list()
train_ratio = 0.7
valid_ratio = 0.1
for t in set(T_uncen_sampling):
UU_temp = UU_uncen_sampling[T_uncen_sampling == t]
U_train.extend(UU_temp[0:int(UU_temp.shape[0]*train_ratio)])
U_valid.extend(UU_temp[int(UU_temp.shape[0]*train_ratio):int(UU_temp.shape[0]*(train_ratio+valid_ratio))])
U_test.extend(UU_temp[int(UU_temp.shape[0]*(train_ratio+valid_ratio)):])
X_temp = X_uncen_sampling[T_uncen_sampling == t]
X_train.extend(X_temp[0:int(X_temp.shape[0]*train_ratio)])
X_valid.extend(X_temp[int(X_temp.shape[0]*train_ratio):int(X_temp.shape[0]*(train_ratio+valid_ratio))])
X_test.extend(X_temp[int(X_temp.shape[0]*(train_ratio+valid_ratio)):])
T_temp = T_uncen_sampling[T_uncen_sampling == t]
T_train.extend(T_temp[0:int(T_temp.shape[0]*train_ratio)])
T_valid.extend(T_temp[int(T_temp.shape[0]*train_ratio):int(T_temp.shape[0]*(train_ratio+valid_ratio))])
T_test.extend(T_temp[int(T_temp.shape[0]*(train_ratio+valid_ratio)):])
C_temp = C_uncen_sampling[T_uncen_sampling == t]
C_train.extend(C_temp[0:int(C_temp.shape[0]*train_ratio)])
C_valid.extend(C_temp[int(C_temp.shape[0]*train_ratio):int(C_temp.shape[0]*(train_ratio+valid_ratio))])
C_test.extend(C_temp[int(C_temp.shape[0]*(train_ratio+valid_ratio)):])
UU_cen, X_cen, T_cen, C_cen = np.array(UU_cen), np.array(X_cen), np.array(T_cen), np.array(C_cen)
s_cen = ran_seed(UU_cen.shape[0])
UU_cen, X_cen, T_cen, C_cen = UU_cen[s_cen], X_cen[s_cen], T_cen[s_cen], C_cen[s_cen]
s_num = UU_uncen_sampling.shape[0]
UU_cen, X_cen, T_cen, C_cen = UU_cen[:s_num], X_cen[:s_num], T_cen[:s_num], C_cen[:s_num]
U_train.extend(UU_cen[0:int(s_num*train_ratio)])
U_valid.extend(UU_cen[int(s_num*train_ratio):int(s_num*(train_ratio+valid_ratio))])
U_test.extend(UU_cen[int(s_num*(train_ratio+valid_ratio)):])
X_train.extend(X_cen[0:int(s_num*train_ratio)])
X_valid.extend(X_cen[int(s_num*train_ratio):int(s_num*(train_ratio+valid_ratio))])
X_test.extend(X_cen[int(s_num*(train_ratio+valid_ratio)):])
T_train.extend(T_cen[0:int(s_num*train_ratio)])
T_valid.extend(T_cen[int(s_num*train_ratio):int(s_num*(train_ratio+valid_ratio))])
T_test.extend(T_cen[int(s_num*(train_ratio+valid_ratio)):])
C_train.extend(C_cen[0:int(s_num*train_ratio)])
C_valid.extend(C_cen[int(s_num*train_ratio):int(s_num*(train_ratio+valid_ratio))])
C_test.extend(C_cen[int(s_num*(train_ratio+valid_ratio)):])
U_train, X_train, T_train, C_train = np.array(U_train), np.array(X_train), np.array(T_train), np.array(C_train)
U_valid, X_valid, T_valid, C_valid = np.array(U_valid), np.array(X_valid), np.array(T_valid), np.array(C_valid)
U_test, X_test, T_test, C_test = np.array(U_test), np.array(X_test), np.array(T_test), np.array(C_test)
# U_train, X_train, T_train, C_train = U_train[0:3000], X_train[0:3000], T_train[0:3000], C_train[0:3000]
# U_valid, X_valid, T_valid, C_valid = U_valid[0:1200], X_valid[0:1200], T_valid[0:1200], C_valid[0:1200]
# U_test, X_test, T_test, C_test = U_test[0:1800], X_test[0:1800],T_test[0:1800], C_test[0:1800]
# print U_train.shape, U_valid.shape, U_test.shape
# print X_train.shape, X_valid.shape, X_test.shape
# print T_train.shape, T_valid.shape, T_test.shape
# print C_train.shape, C_valid.shape, C_test.shape
np.save(dest_path + "U_train",U_train)
np.save(dest_path + "X_train",X_train)
np.save(dest_path + "T_train",T_train)
np.save(dest_path + "C_train",C_train)
np.save(dest_path + "U_valid",U_valid)
np.save(dest_path + "X_valid",X_valid)
np.save(dest_path + "T_valid",T_valid)
np.save(dest_path + "C_valid",C_valid)
np.save(dest_path + "U_test",U_test)
np.save(dest_path + "X_test",X_test)
np.save(dest_path + "T_test",T_test)
np.save(dest_path + "C_test",C_test)
exit(0)