-
Notifications
You must be signed in to change notification settings - Fork 4
/
run.py
162 lines (127 loc) · 6.73 KB
/
run.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
# -*- coding: utf-8 -*-
import os
import numpy as np
import pandas as pd
import tensorflow as tf
from sklearn import metrics
import argparse
from data_factory import dataprocessing
from data_factory import BatchLoader
from configs import BasicParam
from model import HEBR_TSC
def getmetric(y, y_pred):
return {'acc': metrics.accuracy_score(y, y_pred),
'pre': metrics.precision_score(y, y_pred),
're': metrics.recall_score(y, y_pred),
'f1': metrics.f1_score(y, y_pred)}
def splitTrainTest(datax, datay):
def norm(data):
return (data - np.mean(data, axis=0)) / (np.std(data, axis=0) + 1e-20)
df = pd.DataFrame({'time': [pd.to_datetime('2018-11-1')]})
splittamp = (pd.to_datetime(df['time']).astype(int)/ 1e9).values[0]
trainx, trainy, testx, testy = [], [], [], []
for i in range(len(datax)):
x = datax[i][-180:, [1,3,5,6,7,8]] # Total，on-peak，off-peak，ntl，high temperature，low temperature
x = norm(x) # shape: [his_len, dim]
y = datay[i]
if (y[1] >= splittamp) and (y[1] < splittamp+30*24*60*60):
testx.append(x)
testy.append(y[0])
else:
trainx.append(x)
trainy.append(y[0])
trainx = np.asarray(trainx, dtype=np.float32)
trainy = np.asarray(trainy, dtype=np.int32)
testx = np.asarray(testx, dtype=np.float32)
testy = np.asarray(testy, dtype=np.int32)
return trainx, trainy, testx, testy
def getparams():
parser = argparse.ArgumentParser()
parser.add_argument('-f', '--data_file', type=str, default='data/', help='path of input file')
parser.add_argument('-o', '--output_file', type=str, default='data/result.csv', help='path of output file')
parser.add_argument('-hl', '--history_lenght', type=int, default=180,
help='the historical length for observed data, default value is 180')
parser.add_argument('-b', '--batch_size', type=int, default=1000,
help='the number of samples in each batch, default value is 1000')
parser.add_argument('-e', '--num_epoch', type=int, default=100, help='number of epoch, default value is 100')
parser.add_argument('-n', '--cpu_jobs', type=int, default=os.cpu_count(),
help='number of cpu jobs, default value is maximum number of cpu kernel')
parser.add_argument('-g', '--gpu_id', type=int, default=0,
help='index of gpu, default value is 0')
parser.add_argument('-l', '--learning_rate', type=float, default=0.001)
parser.add_argument('-iu', '--user_dims', type=int, default=16,
help='dimension of micro-level memory matrix, default value is 16')
parser.add_argument('-il', '--ntl_dims', type=int, default=4,
help='dimension of meso-level memory matrix, default value is 4')
parser.add_argument('-ie', '--climate_dims', type=int, default=8,
help='dimension of macro-level memory matrix, default value is 8')
parser.add_argument('-iul', '--user_ntl_dims', type=int, default=64,
help='dimension of user-area memory matrix, default value is 64')
parser.add_argument('-iue', '--user_climate_dims', type=int, default=64,
help='dimension of user-climate memory matrix, default value is 64')
parser.add_argument('-iule', '--user_ntl_climate_dims', type=int, default=256,
help='dimension of user_ntl_climate memory matrix, default value is 256')
args = parser.parse_args()
configs = BasicParam()
configs.data_path = args.data_file
configs.his_len = args.history_file
configs.n_jobs = args.cpu_jobs
configs.gpu = str(args.gpu_id)
configs.batch_size = args.batch_size
configs.learning_rate = args.learning_rate
configs.epoch = args.num_epoch
configs.dims['user_hidden'] = args.user_dims
configs.dims['area_hidden'] = args.ntl_dims
configs.dims['climate_hidden'] = args.climate_dims
configs.dims['user_area_hidden'] = args.user_ntl_dims
configs.dims['user_climate_hidden'] = args.user_climate_dims
configs.dims['user_area_climate_hidden'] = args.user_ntl_climate_dims
configs.dims['user_area'] = 2 * configs.dims['user_hidden']
configs.dims['user_climate'] = 2 * configs.dims['user_hidden']
configs.dims['user_area_climate'] = 2 * configs.dims['user_area_hidden']
return configs
if __name__ == '__main__':
configs = getparams()
os.environ["CUDA_VISIBLE_DEVICES"] = configs.gpu
# data processing
dataprocessing(configs.data_path)
# load data
datax = np.load(os.path.join(configs.data_path, 'datax.npy'), allow_pickle=True)
datay = np.load(os.path.join(configs.data_path, 'datay.npy'), allow_pickle=True)
trainx, trainy, testx, testy = splitTrainTest(datax, datay)
print(trainx.shape, trainy.shape, np.sum(trainy), (trainy.shape[0] - np.sum(trainy)) / np.sum(trainy))
print(testx.shape, testy.shape, np.sum(testy), (testy.shape[0] - np.sum(testy)) / np.sum(testy))
# establish dataloader
trainloader = BatchLoader(configs.batch_size)
trainloader.load_data(trainx, trainy, shuffle=True)
testloader = BatchLoader(configs.batch_size)
testloader.load_data(testx, testy, shuffle=False)
model = HEBR_TSC()
model.set_configuration(configs)
gpu_options = tf.GPUOptions(per_process_gpu_memory_fraction=1.0)
config = tf.ConfigProto(gpu_options=gpu_options)
with tf.Session(config=config) as sess:
model.build_model(is_training=True)
init_vars = tf.global_variables_initializer()
sess.run(init_vars)
bestP = 0.0
for i in range(100):
loss = model.fit(sess, trainloader)
y_pred, _ = model.predict(sess, testloader)
results = getmetric(testy, y_pred)
logstr = 'Epochs {:d}, loss {:f}, Accuracy {:f}, Precision {:f}, Recall {:f}, F1 {:f}'.format(i, loss, results['acc'], results['pre'], results['re'], results['f1'])
print(logstr)
p = 1.25 * results['pre'] * results['re'] / (0.25 * results['pre'] + results['re'])
if p > bestP:
model.store(configs.save_path, sess=sess)
bestP = p
print('epoch {} store.'.format(i))
print("model testing...")
tf.reset_default_graph()
with tf.Session(config=tf.ConfigProto(allow_soft_placement=True)) as sess:
model.build_model(is_training=False)
model.restore(configs.save_path, sess=sess)
y_pred, _ = model.predict(sess, testloader)
results = getmetric(testy, y_pred)
print('Accuracy {:f}, Precision {:f}, Recall {:f}, F1 {:f}'.format(results['acc'], results['pre'], results['re'], results['f1']))
print(results['matrix'])