-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdataset.py
181 lines (161 loc) · 8.1 KB
/
dataset.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
import pathlib
import re
import numpy as np
import pandas as pd
from scipy import spatial
from sklearn.preprocessing import QuantileTransformer
class UserFactory:
def __init__(self, users: pd.DataFrame, node_coordinates: np.array, functions: np.array, normalization_factor: int):
self.users = users
self.functions = functions
self.user_function_assignment = {}
self.node_coordinates = node_coordinates
self.kd_tree = spatial.KDTree(node_coordinates)
self.normalization_factor = normalization_factor
def get_position(self, timestamp: float):
filtered_users = self.users[:self.users['start'].searchsorted(timestamp, side='left')]
filtered_users = filtered_users[filtered_users['end'] >= timestamp]
filtered_users['function'] = filtered_users['id'].map(lambda x: self.get_user_function(x))
return filtered_users
def get_user_function(self, user_id: int):
if user_id not in self.user_function_assignment:
self.user_function_assignment[user_id] = np.random.choice(
a=range(len(self.functions)),
p=self.functions,
)
return self.user_function_assignment[user_id]
def get_workload(self, timestamp: float):
user_coordinates = self.get_position(timestamp)[['lat', 'long']].to_numpy()
node_workload = [self.kd_tree.query(user)[1] for user in user_coordinates]
node_ids, value_counts = np.unique(node_workload, return_counts=True)
result = np.array([0] * len(self.node_coordinates))
for node_id, value_count in zip(node_ids, value_counts):
result[node_id] = value_count
return result
def get_user_workload(self, timestamp: float):
df = self.get_position(timestamp)
function_dfs = [y for x, y in df.groupby('function', as_index=False)]
result = np.zeros(shape=(len(self.node_coordinates), len(self.functions)))
for function_df in function_dfs:
function = function_df['function'].values[0]
user_coordinates = function_df[['lat', 'long']].to_numpy()
node_workload = [self.kd_tree.query(user)[1] for user in user_coordinates]
node_ids, value_counts = np.unique(node_workload, return_counts=True)
for node_id, value_count in zip(node_ids, value_counts):
result[node_id][function] = value_count / self.normalization_factor
return result
def get_avg_user_workload_in_interval(self, begin_ts: float, end_ts: float, delta=0.01):
result = None
t_samples = range(int(begin_ts/delta), int(end_ts/delta))
for i in t_samples:
try:
result += self.get_user_workload(i*delta)/len(t_samples)
except:
result = self.get_user_workload(i*delta)/len(t_samples)
return result
class CabspottingUserFactory(UserFactory):
def __init__(self, dataset_dir: str, node_coordinates: np.array, functions: np.array):
self.dataset_dir = pathlib.Path(dataset_dir)
cabs_df = pd.read_csv(self.dataset_dir.joinpath("_cabs.txt"), header=None)
cabs_df.columns = ['row']
cabs = cabs_df['row'].map(lambda x: re.findall('"([^"]*)"', x)[0]).to_list()
cabs_info = []
for i, cab_id in enumerate(cabs):
cab_df = pd.read_csv(self.dataset_dir.joinpath(f"new_{cab_id}.txt"), sep=" ", header=None,
names=['lat', 'long', 'state', 'timestamp'], usecols=['lat', 'long', 'timestamp'])
cab_df['id'] = i
cab_df = cab_df.sort_values('timestamp')
timestamps = cab_df['timestamp'].to_list()
cab_df = cab_df[1:]
cab_df['start'] = timestamps[:-1]
cab_df['end'] = timestamps[1:]
cab_df = cab_df.drop(columns='timestamp')
cabs_info.append(cab_df)
users = pd.concat(cabs_info)
# Normalize towards uniform distribution
cols = ['lat', 'long']
scaler = QuantileTransformer()
users[cols] = scaler.fit_transform(users[cols])
min_time = users['start'].min()
max_time = users['end'].max()
for col in ['start', 'end']:
users[col] = (users[col] - min_time) / (max_time - min_time)
users = users.sort_values(['start', 'end'])
super().__init__(users, node_coordinates, functions, 20)
class TDriveUserFactory(UserFactory):
def __init__(self, dataset_dir: str, node_coordinates: np.array, functions: np.array):
self.dataset_dir = pathlib.Path(dataset_dir)
cabs_info = []
for i in range(1, 10357):
file = self.dataset_dir.joinpath(f"release/taxi_log_2008_by_id/{i}.txt")
if file.stat().st_size == 0:
continue
cab_df = pd.read_csv(file, sep=",", header=None, index_col=0, parse_dates=['timestamp'],
date_format="%Y-%m-%d %H:%M:%S", names=['timestamp', 'lat', 'long'])
cab_df.columns = ['timestamp', 'lat', 'long']
cab_df['id'] = i
# cab_df['timestamp'] = cab_df['timestamp']
cab_df = cab_df.sort_values('timestamp')
timestamps = cab_df['timestamp'].to_list()
cab_df = cab_df[1:]
cab_df['start'] = timestamps[:-1]
cab_df['end'] = timestamps[1:]
cab_df = cab_df.drop(columns='timestamp')
cabs_info.append(cab_df)
users = pd.concat(cabs_info)
# Normalize towards uniform distribution
cols = ['lat', 'long']
scaler = QuantileTransformer()
users[cols] = scaler.fit_transform(users[cols])
min_time = users['start'].min()
max_time = users['end'].max()
for col in ['start', 'end']:
users[col] = (users[col] - min_time) / (max_time - min_time)
users = users.sort_values(['start', 'end'])
super().__init__(users, node_coordinates, functions, 400)
class TelecomUserFactory(UserFactory):
def __init__(self, dataset_dir: str, node_coordinates: np.array, functions: np.array):
self.dataset_dir = pathlib.Path(dataset_dir)
users = pd.read_excel(self.dataset_dir.joinpath("data_6.1~6.15.xlsx"),
usecols=['start time', 'end time', 'latitude', 'longitude', 'user id'],
parse_dates=['start time', 'end time'],
date_format="%Y-%m-%d %H:%M:%S", )
users = users[~(users.isna().sum(axis=1).astype(bool))]
users['user id'] = users['user id'].astype('category').cat.codes
users = users.rename(
columns={'start time': 'start', 'end time': 'end', 'user id': 'id', 'latitude': 'lat', 'longitude': 'long'})
normalized_cabs = (users - users.min()) / (users.max() - users.min())
users['lat'] = normalized_cabs['lat'].astype(float)
users['long'] = normalized_cabs['long'].astype(float)
users['start'] = normalized_cabs['start'].astype(float)
users['end'] = normalized_cabs['end'].astype(float)
# Normalize towards uniform distribution
cols = ['lat', 'long']
scaler = QuantileTransformer()
users[cols] = scaler.fit_transform(users[cols])
users = users.sort_values(['start', 'end'])
super().__init__(users, node_coordinates, functions, 60)
if __name__ == '__main__':
np.random.seed(0)
n = 1000
node_coordinates = np.array([
[0.25, 0.25],
[0.25, 0.75],
[0.75, 0.25],
[0.75, 0.75],
])
functions = np.array([
1, 1, 1, 1, 1, 1, 1, 0.1
]) # probability weights
functions = functions / sum(functions)
#user_factory = CabspottingUserFactory("cabspottingdata", node_coordinates, functions=functions)
# a = [sum(user_factory.get_workload(t / n)) for t in range(1, n)]
# print(min(a))
# print(max(a))
user_factory = TDriveUserFactory("tdrive", node_coordinates, functions)
print(user_factory.get_avg_user_workload_in_interval(0.1, 0.2, 0.01))
# user_factory = TelecomUserFactory("telecom", node_coordinates)
# a = [sum(user_factory.get_workload(t/n)) for t in range(1, n)]
# print(min(a))
# print(max(a))
# pass