-
Notifications
You must be signed in to change notification settings - Fork 3
/
main_mtl_exchange_rate.py
158 lines (128 loc) · 7.08 KB
/
main_mtl_exchange_rate.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
from keras.callbacks import EarlyStopping
import pandas as pd
from common.TimeseriesTensor import TimeSeriesTensor
from common.gp_log import store_training_loss, store_predict_points, flatten_test_predict
from common.utils import load_data, split_train_validation_test, load_data_full, mape
from ts_model import create_model, create_model_mtl_mtv_temperature, \
create_model_mtl_mtv_exchange_rate
from kgp.metrics import root_mean_squared_error as RMSE
import matplotlib.pyplot as plt
from keras.callbacks import ModelCheckpoint
import numpy as np
from sklearn.metrics import explained_variance_score
from sklearn.metrics import mean_absolute_error
from sklearn.metrics import mean_squared_error
from sklearn.metrics import mean_squared_log_error
from sklearn.metrics import median_absolute_error
from sklearn.metrics import r2_score
import os
if __name__ == '__main__':
time_step_lag = 6
HORIZON = 1
imfs_count = 11
data_dir = 'data'
# output_dir = 'output/exchange-rate/mtl/lag' + str(time_step_lag)
output_dir = 'output/exchange-rate/mtl_mtv/horizon_' + str(HORIZON) + '/lag' + str(time_step_lag)
os.makedirs(output_dir, exist_ok=True)
os.makedirs(output_dir + '/model_checkpoint', exist_ok=True)
multi_time_series = load_data_full(data_dir, datasource='exchange-rate', imfs_count=imfs_count, freq='d')
print(multi_time_series.head())
valid_start_dt = '2002-06-18'
test_start_dt = '2006-08-13'
features = ["load", "imf9", "imf10", "imf8", "imf7"]
targets = ["load", "imf9", "imf10", "imf8", "imf7"]
train_inputs, valid_inputs, test_inputs, y_scaler = split_train_validation_test(multi_time_series,
valid_start_time=valid_start_dt,
test_start_time=test_start_dt,
time_step_lag=time_step_lag,
horizon=HORIZON,
features=features,
target=targets,
time_format = '%Y-%m-%d',
freq = 'd'
)
# ['imf6', 'imf5', 'imf4', 'imf3', 'imf2', 'imf0', 'imf1']
aux_features = ["load", "imf6", "imf5", 'imf4', 'imf3', 'imf2', 'imf0', 'imf1']
# for i in range(imfs_count):
# l = 'imf' + str(i)
# if l not in features:
# aux_features.append(l)
aux_inputs, aux_valid_inputs, aux_test_inputs, aux_y_scaler = split_train_validation_test(multi_time_series,
valid_start_time=valid_start_dt,
test_start_time=test_start_dt,
time_step_lag=time_step_lag,
horizon=HORIZON,
features=aux_features,
target=["load"],
time_format = '%Y-%m-%d',
freq = 'd'
)
X_train = train_inputs['X']
y1_train = train_inputs['target_load']
y2_train = train_inputs['target_imf9']
y3_train = train_inputs['target_imf10']
y4_train = train_inputs['target_imf8']
y5_train = train_inputs['target_imf7']
y_train = [y1_train, y2_train, y3_train, y4_train, y5_train]
X_valid = valid_inputs['X']
y1_valid = valid_inputs['target_load']
y2_valid = valid_inputs['target_imf9']
y3_valid = valid_inputs['target_imf10']
y4_valid = valid_inputs['target_imf8']
y5_valid = valid_inputs['target_imf7']
y_valid = [y1_valid, y2_valid, y3_valid, y4_valid, y5_valid]
aux_train = aux_inputs['X']
aux_valid = aux_valid_inputs['X']
aux_test = aux_test_inputs['X']
# input_x = train_inputs['X']
print("train_X shape", X_train.shape, "train_Y shape:", y1_train.shape)
print("valid_X shape", X_valid.shape, "valid Y shape:", y1_valid.shape)
print("aux_train shape", aux_train.shape, "aux valid Y shape", aux_valid.shape)
# print("target shape", y_train.shape)
# print("training size:", len(train_inputs['X']), 'validation', len(valid_inputs['X']), 'test size:', len(test_inputs['X']) )
# print("sum sizes", len(train_inputs['X']) + len(valid_inputs['X']) + len(test_inputs['X']))
# LATENT_DIM = 5
BATCH_SIZE = 32
EPOCHS = 30
model = create_model_mtl_mtv_exchange_rate(horizon=HORIZON, nb_train_samples=len(X_train),
batch_size=32, feature_count=len(features), lag_time=time_step_lag,
aux_feature_count=len(aux_features))
earlystop = EarlyStopping(monitor='loss', patience=5)
file_path = output_dir + '/model_checkpoint/weights-improvement-{epoch:02d}.hdf5'
check_point = ModelCheckpoint(file_path, monitor='val_loss', verbose=0, save_best_only=True,
save_weights_only=True, mode='auto', period=1)
history = model.fit([X_train, aux_train],
y_train,
batch_size=BATCH_SIZE,
epochs=EPOCHS,
validation_data=([X_valid, aux_valid], y_valid),
callbacks=[earlystop, check_point],
verbose=1)
store_training_loss(history=history, filepath=output_dir + "/training_loss_epochs_" + str(EPOCHS) + "_lag" +
str(time_step_lag) + ".csv")
# Finetune the model
# model.finetune(X_train, y_train, batch_size=BATCH_SIZE, gp_n_iter=10, verbose=1)
# Test the model
X_test = test_inputs['X']
y1_test = test_inputs['target_load']
y2_test = test_inputs['target_imf9']
y3_test = test_inputs['target_imf10']
y4_test = test_inputs['target_imf8']
y5_test = test_inputs['target_imf7']
y1_preds, y2_preds, y3_preds, y4_preds, y5_preds = model.predict([X_test, aux_test])
# y1_preds, y2_preds, y3_preds, y4_preds = model.predict([X_test, aux_test])
y1_test = y_scaler.inverse_transform(y1_test)
y1_preds = y_scaler.inverse_transform(y1_preds)
y1_test, y1_preds = flatten_test_predict(y1_test, y1_preds)
rmse_predict = RMSE(y1_test, y1_preds)
evs = explained_variance_score(y1_test, y1_preds)
mae = mean_absolute_error(y1_test, y1_preds)
mse = mean_squared_error(y1_test, y1_preds)
msle = mean_squared_log_error(y1_test, y1_preds)
meae = median_absolute_error(y1_test, y1_preds)
r_square = r2_score(y1_test, y1_preds)
mape_v = mape(y1_preds.reshape(-1, 1), y1_test.reshape(-1, 1))
print('rmse_predict:', rmse_predict, "evs:", evs, "mae:", mae,
"mse:", mse, "msle:", msle, "meae:", meae, "r2:", r_square, "mape", mape_v)
store_predict_points(y1_test, y1_preds, output_dir + '/test_mtl_prediction_epochs_' + str(EPOCHS) + '_lag_'
+ str(time_step_lag) + '.csv')