-
Notifications
You must be signed in to change notification settings - Fork 13
/
params_stg2seq.py
121 lines (116 loc) · 2.47 KB
/
params_stg2seq.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
class params_SY(object):
#commen params
source = 'SY'
scaler = 81
batch_size = 32
map_height = 6
map_width = 6
closeness_sequence_length = 8
period_sequence_length = 0
trend_sequence_length = 0
nb_flow = 1
external_length = 8
et_dim = 33
ew_dim = 7
horizon = 3
lr = 0.0001
num_epochs = 400
logdir = "train"
test_days = 6
delta = 0.5
epsilon = 1e-7
beta1 = 0.8
beta2 = 0.999
#specific params
filter_type = 'random_walk'
num_layers = 2
num_units = 64
num_nodes = map_height*map_width
keep_prob = 1
exlstm_layers = 5
class params_SY_IR(object):
#when DidiSY dataset is partitioned by roadnetworks
source = 'SY_IR'
scaler = 70
batch_size = 32
map_height = 35
map_width = 1
closeness_sequence_length = 8
period_sequence_length = 0
trend_sequence_length = 0
nb_flow = 1
external_length = 8
et_dim = 33
ew_dim = 7
horizon = 3
lr = 0.0005
num_epochs = 1200
logdir = "train"
test_days = 6
delta = 0.5
epsilon = 1e-7
beta1 = 0.8
beta2 = 0.999
#specific params
filter_type = 'random_walk'
num_layers = 2
num_units = 64
num_nodes = map_height*map_width
keep_prob = 1
exlstm_layers = 5
class params_NYC(object):
source = 'NYC'
scaler = 267
batch_size = 32
map_height = 16
map_width = 8
closeness_sequence_length = 5
period_sequence_length = 0
trend_sequence_length = 0
nb_flow = 2
external_length = closeness_sequence_length
et_dim = 31
ew_dim = 0
horizon = 3
lr = 0.0007
num_epochs = 150
logdir = "train"
test_days = 10
delta = 0.5
epsilon = 1e-7
beta1 = 0.8
beta2 = 0.999
filter_type = 'random_walk'
num_layers = 2
num_units = 64
num_nodes = map_height * map_width
keep_prob = 1
exlstm_layers = 3
class params_BJ(object):
source = 'BJ'
scaler = 1274
batch_size = 32
map_height = 32
map_width = 32
closeness_sequence_length = 8
period_sequence_length = 0
trend_sequence_length = 0
nb_flow = 2
external_length = 8
et_dim = 57
ew_dim = 19
horizon = 3
lr = 0.001
num_epochs = 300
logdir = "train"
test_days = 10
delta = 0.5
epsilon = 1e-7
beta1 = 0.8
beta2 = 0.999
filter_type = 'random_walk'
num_layers = 2
num_units = 64
num_nodes = map_height * map_width
keep_prob = 1
exlstm_layers = 2