-
Notifications
You must be signed in to change notification settings - Fork 0
/
rnn_helper.py
155 lines (110 loc) · 4.34 KB
/
rnn_helper.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
import numpy as np
def generate_batches_regression(data, batch_size, learn_window, predict_window,
start_days=[], seed=None):
"""
Randomly generate batches for regression. Returns Generator that
yields tuples (X, y), where X contains the data to learn and y
contains the data to predict.
- data: shape = (N, C)
- batch_size: integer
- learn_window: array of indices to learn from.
- predict_window: array of indices to predict.
- start_days: array of indices that we can start a learn window on. If empty, then uses entire data.
- seed: seed for RNG, if any.
e.g. if learn_window = [0, 1, 2, ..., 9] and predict_window = [0, 1, 2, 3, 4],
then learn on 10 days and try to predict next 5 days.
returns: generator that yields a tuple (X, y)
- X: shape = (batch_size, len(learn_window), C)
- y: shape = (batch_size, len(predict_window))
"""
if seed != None:
np.random.seed(seed)
learn_window = np.array(learn_window)
predict_window = np.array(predict_window)
N, C = data.shape
L, P = len(learn_window), len(predict_window)
Lcap, Pcap = int(np.max(learn_window))+1, int(np.max(predict_window))+1
M = 0 # size of start_days
if len(start_days) == 0:
M = N - Lcap - Pcap
start_days = np.arange(M)
else:
M = len(start_days)
X = np.zeros((batch_size, L, C))
y = np.zeros((batch_size, P))
i = 0 # tracks index
k = 0 # tracks batch size
while True:
i = start_days[np.random.randint(0, M)]
X[k] = data[i+learn_window]
y[k] = data[i+Lcap+predict_window, 0] ## USD-EUR Close is index 0
k += 1
if k == batch_size:
k = 0
yield X, y
def generate_all_regression(data, batch_size, learn_window, predict_window, start_days=[]):
"""
Deterministic version of generate_batches_regression.
"""
learn_window = np.array(learn_window)
predict_window = np.array(predict_window)
N, C = data.shape
L, P = len(learn_window), len(predict_window)
Lcap, Pcap = int(np.max(learn_window))+1, int(np.max(predict_window))+1
M = 0 # size of start_days
if len(start_days) == 0:
M = N - Lcap - Pcap
start_days = np.arange(M)
else:
M = len(start_days)
X = np.zeros((batch_size, L, C))
y = np.zeros((batch_size, P))
k = 0 # tracks batch size
for j in range(M):
i = start_days[j]
X[k] = data[i+learn_window]
y[k] = data[i+Lcap+predict_window, 0] ## USD-EUR Close is index 0
k += 1
if k == batch_size:
k = 0
yield X, y
def generate_batches_accuracy(data, batch_size, learn_window, seed=None):
"""
Randomly generate batches for accuracy. Here y contains 1 if next close
value is higher, or 0 if next close value is lower.
returns: generator that yields a tuple (X, y)
- X: shape = (batch_size, len(learn_window), C)
- y: shape = (batch_size, 1)
"""
if seed != None:
np.random.seed(seed)
learn_window = np.array(learn_window)
N, C = data.shape
L, P = len(learn_window), 1
Lcap, Pcap = int(np.max(learn_window))+1, 1
X = np.zeros((batch_size, L, C))
y = np.zeros((batch_size, P))
i = 0 # tracks index
k = 0 # tracks batch size
M = N - Lcap - Pcap # ceiling for i
while True:
if k == batch_size:
k = 0
yield X, y
i = np.random.randint(0, M)
X[k] = data[i+learn_window]
y[k, 0] = data[i+Lcap-1, 0] < data[i+Lcap, 0] ## USD-EUR Close is index 0
k += 1
def test_rnn_helper():
data = np.arange(1000).reshape(-1, 1)
print("generate_batches_regression...")
G = generate_batches_regression(data, 3, np.arange(10), np.arange(5))
X, y = next(G)
for i in range(3):
print(i, "X:", X[i, :, 0])
print(i, "y:", y[i, :])
print("get_all_regression...")
X, y = get_all_regression(data, np.arange(10), np.arange(5), [0, 100, 200])
for i in range(3):
print(i, "X:", X[i, :, 0])
print(i, "y:", y[i, :])