-
Notifications
You must be signed in to change notification settings - Fork 0
/
myDataset.py
251 lines (213 loc) · 7.79 KB
/
myDataset.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
#!/usr/bin/env python
# coding=utf-8
import os
import torch
import setproctitle
setproctitle.setproctitle("cvae@s")
import numpy as np
import json
import pickle
import torch
import pdb
import os
from utils import merge,identify_home_
class myDataset_mobile(object):
def __init__(self, split:str, control:str):
super().__init__()
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
self.max_grid = 180
self.split_data_dir = '/data/zmy/dataset/mobile/split_label'
# self.split_data_dir = '/data/zmy/dataset/tencent/split_label'
self.max_pos = 168
self.label_type_num = 1
self.seq_len = self.max_pos + self.label_type_num + 4
self.control = control
self.read_split_data(split)
def read_split_data(self,split:str):
path = os.path.join(self.split_data_dir, split+'.pkl')
self.split_data = pickle.load(open(path, 'rb'))
print(split+" data loaded...")
def __getitem__(self, id):
item = np.zeros((self.max_pos, 3), dtype=int) # (169,3)
idx = 0
traj_data = self.split_data[id]['traj'] # mobile
revenue, gender, edu, age = self.split_data[id]['profile'] # mobile
# traj_data = self.split_data[id] # tencent
'''traj信息'''
for j in range(len(traj_data)): # 天数*24
for i in range(24):
if idx < self.max_pos + 1:
x = traj_data[j][i][0]
y = traj_data[j][i][1]
item[idx][0] = x * self.max_grid +y +1 # grid id
item[idx][1] = i + 1 # hour
item[idx][2] = traj_data[j][i][2] +1 # y # week
idx += 1
else:
break
'''home实验'''
traj = item[:,0,...].copy() # traj (168,)
label = {}
if self.control == 'home':
traj_temp = merge(traj)
home = identify_home_(traj_temp)
label['home'] = home
elif self.control == 'revenue':
if revenue<40:
revenue_label = 1 # 32392
elif revenue<80:
revenue_label = 2
elif revenue<110:
revenue_label = 3
else:
revenue_label = 4
revenue_label += 32391
label['revenue'] = revenue_label
elif self.control == 'edu':
if edu.startswith('初中'):
edu_label = 1
elif edu.startswith('高中'):
edu_label = 2
elif edu.startswith('本科'):
edu_label = 3
else:
edu_label = 4
edu_label += 32391
label['edu'] = edu_label
elif self.control == 'age':
if age<30:
age_label = 1
elif age<40:
age_label = 2
elif age<60:
age_label = 3
else:
age_label = 4
age_label += 32391
label['age'] = age_label
elif self.control == 'gender':
if int(gender)==1:
gender_label = 1
else:
gender_label = 2
gender_label += 32391
label['gender'] = gender_label
labels = np.zeros((self.label_type_num,), dtype=int) # (1,)
labels[-1] = label[self.control]
token = np.array([32411], dtype=int)
start_token = np.array([32412,32412], dtype=int)
x = np.concatenate([start_token,labels,token], axis=0)
y = np.concatenate([start_token,labels,token,traj,token], axis=0)
input = np.concatenate([token,start_token,labels,token,traj], axis=0)
# mask
x_mask = torch.full((self.label_type_num+1+2, ), 1).to(self.device)
y_mask = torch.full((self.seq_len, ), 1).to(self.device)
x = torch.tensor(x).to(self.device)
y = torch.tensor(y).to(self.device)
input = torch.tensor(input).to(self.device)
dict = {"x_mask":x_mask,
"x_tokens":x,
"y_mask":y_mask,
"y_tokens":y,
"input_tokens":input,
"target_tokens":y,
"mask":y_mask}
return dict
def __len__(self):
return len(self.split_data)
class myDataset_age(object):
def __init__(self, split:str):
super().__init__()
self.device = torch.device("cuda" if torch.cuda.is_available() else "cpu")
self.max_grid = 180
self.split_data_dir = '/data/zmy/dataset/chinamobile/split_label'
# self.split_data_dir = '/data/zmy/dataset/tencent/split'
self.max_pos = 168
self.label_type_num = 5
self.seq_len = self.max_pos + self.label_type_num + 2
self.read_split_data(split)
def read_split_data(self,split:str):
path = os.path.join(self.split_data_dir, split+'.pkl')
self.split_data = pickle.load(open(path, 'rb'))
print(split+" data loaded...")
def __getitem__(self, id):
item = np.zeros((self.max_pos+1, 3), dtype=int) # (169,3)
idx = 1 # 0为start token
traj_data = self.split_data[id]['traj'] # mobile
revenue, gender, edu, age = self.split_data[id]['profile'] # mobile
# traj_data = self.split_data[id] # tencent
'''traj信息'''
for j in range(len(traj_data)): # 天数*24
for i in range(24):
if idx < self.max_pos + 1:
x = traj_data[j][i][0]
y = traj_data[j][i][1]
item[idx][0] = x * self.max_grid +y +1 # grid id
item[idx][1] = i +1 # hour
item[idx][2] = traj_data[j][i][2] +1 # y # week
idx += 1
else:
break
'''home实验'''
# this is for home control
traj = item[:,0,...].copy() # traj (169,)
traj_temp = merge(traj[1:])
home = identify_home_(traj_temp)
labels = np.zeros((self.label_type_num,), dtype=int) # (5,)
pdb.set_trace()
labels[-1] = home # home
# mask
mask_long = torch.full((self.seq_len, ), 1).to(self.device)
mask_short = torch.full((self.label_type_num+1, ), 1).to(self.device)
traj = np.append(traj,int(home)) # 带label的traj (170,)
labels = traj.copy()
labels[1:self.max_pos+1] = -100
traj = torch.tensor(traj)
traj = traj.to(self.device)
labels = torch.tensor(labels)
labels = labels.to(self.device)
'''
# profile信息
if revenue<50:
revenue_label = 1
elif revenue<100:
revenue_label = 2
elif revenue<150:
revenue_label = 3
elif revenue<200:
revenue_label = 4
else:
revenue_label = 5
if edu.startswith('初中'):
edu_label = 1
elif edu.startswith('高中'):
edu_label = 2
elif edu.startswith('本科'):
edu_label = 3
elif edu.startswith('研究生'):
edu_label = 4
else:
raise ValueError("edu error")
if age<20:
age_label = 1
elif age<30:
age_label = 2
elif age<40:
age_label = 3
elif age<50:
age_label = 4
elif age<60:
age_label = 5
else:
age_label = 6
'''
# return {"data":item,
# "length":len(traj_data),
# "age_label":age_label,
# "revenue_label":revenue_label,
# "gender_label":int(gender),
# "edu_label":edu_label}
return {"input_ids":traj,
"labels":labels}
def __len__(self):
return len(self.split_data)