-
Notifications
You must be signed in to change notification settings - Fork 55
/
sparse_traces.py
284 lines (255 loc) · 12.3 KB
/
sparse_traces.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
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
from __future__ import print_function
from __future__ import division
import time
import argparse
import numpy as np
import cPickle as pickle
from collections import Counter
def entropy_spatial(sessions):
locations = {}
days = sorted(sessions.keys())
for d in days:
session = sessions[d]
for s in session:
if s[0] not in locations:
locations[s[0]] = 1
else:
locations[s[0]] += 1
frequency = np.array([locations[loc] for loc in locations])
frequency = frequency / np.sum(frequency)
entropy = - np.sum(frequency * np.log(frequency))
return entropy
class DataFoursquare(object):
def __init__(self, trace_min=10, global_visit=10, hour_gap=72, min_gap=10, session_min=2, session_max=10,
sessions_min=2, train_split=0.8, embedding_len=50):
tmp_path = "../data/"
self.TWITTER_PATH = tmp_path + 'foursquare/tweets_clean.txt'
self.VENUES_PATH = tmp_path + 'foursquare/venues_all.txt'
self.SAVE_PATH = tmp_path
self.save_name = 'foursquare'
self.trace_len_min = trace_min
self.location_global_visit_min = global_visit
self.hour_gap = hour_gap
self.min_gap = min_gap
self.session_max = session_max
self.filter_short_session = session_min
self.sessions_count_min = sessions_min
self.words_embeddings_len = embedding_len
self.train_split = train_split
self.data = {}
self.venues = {}
self.words_original = []
self.words_lens = []
self.dictionary = dict()
self.words_dict = None
self.data_filter = {}
self.user_filter3 = None
self.uid_list = {}
self.vid_list = {'unk': [0, -1]}
self.vid_list_lookup = {}
self.vid_lookup = {}
self.pid_loc_lat = {}
self.data_neural = {}
# ############# 1. read trajectory data from twitters
def load_trajectory_from_tweets(self):
with open(self.TWITTER_PATH) as fid:
for i, line in enumerate(fid):
_, uid, _, _, tim, _, _, tweet, pid = line.strip('\r\n').split('')
if uid not in self.data:
self.data[uid] = [[pid, tim]]
else:
self.data[uid].append([pid, tim])
if pid not in self.venues:
self.venues[pid] = 1
else:
self.venues[pid] += 1
# ########### 3.0 basically filter users based on visit length and other statistics
def filter_users_by_length(self):
uid_3 = [x for x in self.data if len(self.data[x]) > self.trace_len_min]
pick3 = sorted([(x, len(self.data[x])) for x in uid_3], key=lambda x: x[1], reverse=True)
pid_3 = [x for x in self.venues if self.venues[x] > self.location_global_visit_min]
pid_pic3 = sorted([(x, self.venues[x]) for x in pid_3], key=lambda x: x[1], reverse=True)
pid_3 = dict(pid_pic3)
session_len_list = []
for u in pick3:
uid = u[0]
info = self.data[uid]
topk = Counter([x[0] for x in info]).most_common()
topk1 = [x[0] for x in topk if x[1] > 1]
sessions = {}
for i, record in enumerate(info):
poi, tmd = record
try:
tid = int(time.mktime(time.strptime(tmd, "%Y-%m-%d %H:%M:%S")))
except Exception as e:
print('error:{}'.format(e))
continue
sid = len(sessions)
if poi not in pid_3 and poi not in topk1:
# if poi not in topk1:
continue
if i == 0 or len(sessions) == 0:
sessions[sid] = [record]
else:
if (tid - last_tid) / 3600 > self.hour_gap or len(sessions[sid - 1]) > self.session_max:
sessions[sid] = [record]
elif (tid - last_tid) / 60 > self.min_gap:
sessions[sid - 1].append(record)
else:
pass
last_tid = tid
sessions_filter = {}
for s in sessions:
if len(sessions[s]) >= self.filter_short_session:
sessions_filter[len(sessions_filter)] = sessions[s]
session_len_list.append(len(sessions[s]))
if len(sessions_filter) >= self.sessions_count_min:
self.data_filter[uid] = {'sessions_count': len(sessions_filter), 'topk_count': len(topk), 'topk': topk,
'sessions': sessions_filter, 'raw_sessions': sessions}
self.user_filter3 = [x for x in self.data_filter if
self.data_filter[x]['sessions_count'] >= self.sessions_count_min]
# ########### 4. build dictionary for users and location
def build_users_locations_dict(self):
for u in self.user_filter3:
sessions = self.data_filter[u]['sessions']
if u not in self.uid_list:
self.uid_list[u] = [len(self.uid_list), len(sessions)]
for sid in sessions:
poi = [p[0] for p in sessions[sid]]
for p in poi:
if p not in self.vid_list:
self.vid_list_lookup[len(self.vid_list)] = p
self.vid_list[p] = [len(self.vid_list), 1]
else:
self.vid_list[p][1] += 1
# support for radius of gyration
def load_venues(self):
with open(self.TWITTER_PATH, 'r') as fid:
for line in fid:
_, uid, lon, lat, tim, _, _, tweet, pid = line.strip('\r\n').split('')
self.pid_loc_lat[pid] = [float(lon), float(lat)]
def venues_lookup(self):
for vid in self.vid_list_lookup:
pid = self.vid_list_lookup[vid]
lon_lat = self.pid_loc_lat[pid]
self.vid_lookup[vid] = lon_lat
# ########## 5.0 prepare training data for neural network
@staticmethod
def tid_list(tmd):
tm = time.strptime(tmd, "%Y-%m-%d %H:%M:%S")
tid = tm.tm_wday * 24 + tm.tm_hour
return tid
@staticmethod
def tid_list_48(tmd):
tm = time.strptime(tmd, "%Y-%m-%d %H:%M:%S")
if tm.tm_wday in [0, 1, 2, 3, 4]:
tid = tm.tm_hour
else:
tid = tm.tm_hour + 24
return tid
def prepare_neural_data(self):
for u in self.uid_list:
sessions = self.data_filter[u]['sessions']
sessions_tran = {}
sessions_id = []
for sid in sessions:
sessions_tran[sid] = [[self.vid_list[p[0]][0], self.tid_list_48(p[1])] for p in
sessions[sid]]
sessions_id.append(sid)
split_id = int(np.floor(self.train_split * len(sessions_id)))
train_id = sessions_id[:split_id]
test_id = sessions_id[split_id:]
pred_len = sum([len(sessions_tran[i]) - 1 for i in train_id])
valid_len = sum([len(sessions_tran[i]) - 1 for i in test_id])
train_loc = {}
for i in train_id:
for sess in sessions_tran[i]:
if sess[0] in train_loc:
train_loc[sess[0]] += 1
else:
train_loc[sess[0]] = 1
# calculate entropy
entropy = entropy_spatial(sessions)
# calculate location ratio
train_location = []
for i in train_id:
train_location.extend([s[0] for s in sessions[i]])
train_location_set = set(train_location)
test_location = []
for i in test_id:
test_location.extend([s[0] for s in sessions[i]])
test_location_set = set(test_location)
whole_location = train_location_set | test_location_set
test_unique = whole_location - train_location_set
location_ratio = len(test_unique) / len(whole_location)
# calculate radius of gyration
lon_lat = []
for pid in train_location:
try:
lon_lat.append(self.pid_loc_lat[pid])
except:
print(pid)
print('error')
lon_lat = np.array(lon_lat)
center = np.mean(lon_lat, axis=0, keepdims=True)
center = np.repeat(center, axis=0, repeats=len(lon_lat))
rg = np.sqrt(np.mean(np.sum((lon_lat - center) ** 2, axis=1, keepdims=True), axis=0))[0]
self.data_neural[self.uid_list[u][0]] = {'sessions': sessions_tran, 'train': train_id, 'test': test_id,
'pred_len': pred_len, 'valid_len': valid_len,
'train_loc': train_loc, 'explore': location_ratio,
'entropy': entropy, 'rg': rg}
# ############# 6. save variables
def get_parameters(self):
parameters = {}
parameters['TWITTER_PATH'] = self.TWITTER_PATH
parameters['SAVE_PATH'] = self.SAVE_PATH
parameters['trace_len_min'] = self.trace_len_min
parameters['location_global_visit_min'] = self.location_global_visit_min
parameters['hour_gap'] = self.hour_gap
parameters['min_gap'] = self.min_gap
parameters['session_max'] = self.session_max
parameters['filter_short_session'] = self.filter_short_session
parameters['sessions_min'] = self.sessions_count_min
parameters['train_split'] = self.train_split
return parameters
def save_variables(self):
foursquare_dataset = {'data_neural': self.data_neural, 'vid_list': self.vid_list, 'uid_list': self.uid_list,
'parameters': self.get_parameters(), 'data_filter': self.data_filter,
'vid_lookup': self.vid_lookup}
pickle.dump(foursquare_dataset, open(self.SAVE_PATH + self.save_name + '.pk', 'wb'))
def parse_args():
parser = argparse.ArgumentParser()
parser.add_argument('--trace_min', type=int, default=10, help="raw trace length filter threshold")
parser.add_argument('--global_visit', type=int, default=10, help="location global visit threshold")
parser.add_argument('--hour_gap', type=int, default=72, help="maximum interval of two trajectory points")
parser.add_argument('--min_gap', type=int, default=10, help="minimum interval of two trajectory points")
parser.add_argument('--session_max', type=int, default=10, help="control the length of session not too long")
parser.add_argument('--session_min', type=int, default=5, help="control the length of session not too short")
parser.add_argument('--sessions_min', type=int, default=5, help="the minimum amount of the good user's sessions")
parser.add_argument('--train_split', type=float, default=0.8, help="train/test ratio")
return parser.parse_args()
if __name__ == '__main__':
args = parse_args()
data_generator = DataFoursquare(trace_min=args.trace_min, global_visit=args.global_visit,
hour_gap=args.hour_gap, min_gap=args.min_gap,
session_min=args.session_min, session_max=args.session_max,
sessions_min=args.sessions_min, train_split=args.train_split)
parameters = data_generator.get_parameters()
print('############PARAMETER SETTINGS:\n' + '\n'.join([p + ':' + str(parameters[p]) for p in parameters]))
print('############START PROCESSING:')
print('load trajectory from {}'.format(data_generator.TWITTER_PATH))
data_generator.load_trajectory_from_tweets()
print('filter users')
data_generator.filter_users_by_length()
print('build users/locations dictionary')
data_generator.build_users_locations_dict()
data_generator.load_venues()
data_generator.venues_lookup()
print('prepare data for neural network')
data_generator.prepare_neural_data()
print('save prepared data')
data_generator.save_variables()
print('raw users:{} raw locations:{}'.format(
len(data_generator.data), len(data_generator.venues)))
print('final users:{} final locations:{}'.format(
len(data_generator.data_neural), len(data_generator.vid_list)))