-
Notifications
You must be signed in to change notification settings - Fork 8
/
Copy pathcnn.py
163 lines (91 loc) · 4.33 KB
/
cnn.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
import numpy as np
import util
import tensorflow as tf
from tensorflow import keras
from tensorflow.keras.models import Model
from tensorflow.keras.layers import Layer, Flatten, LeakyReLU
from tensorflow.keras.layers import Input, Reshape, Dense, Lambda
from tensorflow.keras.layers import Conv2D, MaxPooling2D, UpSampling2D
from tensorflow.keras.layers import Conv1D, UpSampling1D
from tensorflow.keras.layers import AveragePooling1D, MaxPooling1D
from tensorflow.keras.layers import LSTM, Dropout
from tensorflow.keras import backend as K
from tensorflow.keras.optimizers import Adam, SGD, RMSprop
from tensorflow.keras.losses import mse, binary_crossentropy
from tensorflow.keras import regularizers, activations, initializers, constraints
from tensorflow.keras.constraints import Constraint
from tensorflow.keras.callbacks import History, EarlyStopping
from tensorflow.keras.utils import plot_model
from tensorflow.keras.models import load_model
from tensorflow.python.keras.utils.generic_utils import get_custom_objects
import string
import numpy as np
import matplotlib as mpl
import matplotlib.pyplot as plt
import matplotlib.gridspec as gridspec
from tqdm import tqdm
import matplotlib.cm as cm
from matplotlib.colors import Normalize
def RMSE(x, y):
return np.sqrt(np.mean(np.square(x.flatten() - y.flatten())))
class CNN:
def __init__(self, data, n_lag, n_seq, name=[]):
self.name = name
self.data = data
self.n_lag = n_lag
self.n_seq = n_seq
self.n_batch = 0
self.nb_epoch = 0
self.x = []
self.y = []
self.model = []
def get_cnn_model(self, input, output):
input_data = Input(shape=(input.shape[1], input.shape[2]))
_ = Conv1D(16, 3, padding='same', data_format='channels_last')(input_data)
_ = LeakyReLU(alpha=0.3)(_)
_ = MaxPooling1D(2, padding="same")(_)
_ = Conv1D(8, 6, padding='same', data_format='channels_last')(_)
_ = LeakyReLU(alpha=0.3)(_)
_ = MaxPooling1D(2, padding="same")(_)
_ = Flatten()(_)
_ = Dropout(0.3)(_)
output_data = Dense(output.shape[1])(_)
self.model = Model(input_data, output_data)
self.model.compile(loss='mse', optimizer=Adam(lr=1e-3, decay=1e-3 / 200))
self.model.summary()
plot_model(self.model, to_file='cnn.png')
def train(self, n_batch=1, nb_epoch=100, load=False):
self.n_batch = n_batch
self.nb_epoch = nb_epoch
x = self.data.iloc[:, 0:self.n_lag].values
self.x = x.reshape(x.shape[0], x.shape[1], 1)
self.y = self.data.iloc[:, self.n_lag:].values
self.get_cnn_model(self.x, self.y)
history = History()
losses = np.zeros([self.nb_epoch, 2])
for i in tqdm(range(self.nb_epoch)):
self.model.fit(self.x, self.y,
epochs=1, batch_size=self.n_batch, verbose=False,
shuffle=True, callbacks=[history])
losses[i, :] = np.asarray(list(history.history.values()))[:, i]
print ("%d [Loss: %f] [Val loss: %f]" % (i, losses[i, 0], losses[i, 1]))
figs = util.plotLosses(losses, name="cnn_losses")
def forecasts(self, test_data):
forecasts = list()
for i in range(len(test_data)):
x = test_data.iloc[i, 0:self.n_lag].values
x = x.reshape(1, len(x), 1)
y = test_data.iloc[i, self.n_lag:].values
y_hat = self.model.predict(x, batch_size=self.n_batch)
forecasts.append([x for x in y_hat[0, :]])
return forecasts
def forecasts_multistep(self, test_data, n_steps=3):
n_steps = n_steps
forecasts_multi = list()
x = test_data.iloc[0, 0:self.n_lag].values
x = x.reshape(1, len(x), 1)
for i in range(n_steps):
y_hat = self.model.predict(x, batch_size=self.n_batch)
forecasts_multi.append(y_hat[0, :])
x = y_hat.reshape(1, y_hat.shape[1], 1)
return forecasts_multi