-
Notifications
You must be signed in to change notification settings - Fork 3
/
utils.py
428 lines (338 loc) · 14.6 KB
/
utils.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
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
import logging
import logging.handlers
import os
from datetime import datetime, date
from pandas.tseries.holiday import USFederalHolidayCalendar as calendar
import matplotlib.pyplot as plt
import pandas as pd
from sklearn.preprocessing import StandardScaler
import numpy as np
import torch
from torch.utils.data import Dataset, DataLoader
from torch.utils.data.sampler import Sampler
def setup_log(subName='', tag='root'):
# create logger
logger = logging.getLogger(tag)
# logger.handlers = []
logger.propagate = False
logger.setLevel(logging.DEBUG)
# create console handler and set level to debug
ch = logging.StreamHandler()
ch.setLevel(logging.DEBUG)
# create formatter
formatter = logging.Formatter('%(asctime)s - %(name)s - %(levelname)s - %(message)s')
# add formatter to ch
ch.setFormatter(formatter)
# add ch to logger
# logger.handlers = []
logger.addHandler(ch)
# file log
log_name = tag + datetime.now().strftime('log_%Y_%m_%d.log')
log_path = os.path.join('log', subName, log_name)
fh = logging.handlers.RotatingFileHandler(
log_path, mode='a', maxBytes=100 * 1024 * 1024, backupCount=1, encoding='utf-8'
)
fh.setFormatter(formatter)
fh.setLevel(logging.DEBUG)
logger.setLevel(logging.DEBUG)
logger.addHandler(fh)
return logger
def save_or_show_plot(file_nm: str, save: bool):
if save:
plt.savefig(os.path.join(os.path.dirname(__file__), "plots", file_nm))
else:
plt.show()
def split_data(x):
return x.iloc[:-365*24], x.iloc[-365*24:]
def get_season(date_time):
# dummy leap year to include leap days(year-02-29) in our range
leap_year = 2000
seasons = [('winter', (date(leap_year, 1, 1), date(leap_year, 3, 20))),
('spring', (date(leap_year, 3, 21), date(leap_year, 6, 20))),
('summer', (date(leap_year, 6, 21), date(leap_year, 9, 22))),
('autumn', (date(leap_year, 9, 23), date(leap_year, 12, 20))),
('winter', (date(leap_year, 12, 21), date(leap_year, 12, 31)))]
if isinstance(date_time, datetime):
date_time = date_time.date()
# we don't really care about the actual year so replace it with our dummy leap_year
date_time = date_time.replace(year=leap_year)
# return season our date falls in.
return next(season for season, (start, end) in seasons
if start <= date_time <= end)
def create_datetype_column(data_set):
# cloning the input dataset.
local = data_set.copy()
# add season column
local['Season'] = pd.Series(local.index).apply(get_season).values
# add holiday column
cal = calendar()
holidays = cal.holidays(start='2008-01-01', end='2020-07-31')
dateTime = pd.to_datetime(data_set.index.date)
local['Holiday'] = dateTime.isin(holidays).astype(int)
# add day of week column
local['day_of_week'] = pd.Series(local.index).dt.day_name().values
# local['day_of_week'] = pd.Series(local.index).dt.dayofweek.values
# add month of year column
local['month_of_year'] = pd.Series(local.index).dt.month_name().values
# local['month_of_year'] = pd.Series(local.index).dt.month.values
# add hour of day column
local['hour_of_day'] = pd.Series(local.index).dt.hour.values
# one-hot encoding
local = pd.get_dummies(local)
local = pd.get_dummies(local, columns=['hour_of_day'])
return local
def mkdir(dirName):
if not os.path.exists(dirName):
if os.name == 'nt':
os.system('mkdir {}'.format(dirName.replace('/', '\\')))
else:
os.system('mkdir -p {}'.format(dirName))
def mkdirectory(config, subName, saveModel):
log_name = '_' + config.logname + datetime.now().strftime('log_%Y_%m_%d')
dirName_data = "data/" + subName
dirName_log = "log/" + subName
mkdir(dirName_data)
mkdir(dirName_log)
if saveModel is True:
model_name = "/model_iso_" + "D_" + str(int(config.past_T / 24)) + "_batch_" + str(
config.batch_size) + "_ed_" + str(config.hidden_size) + "_epochs_" + str(
config.epochs) + log_name
dirName_model = "history_model/" + subName + model_name
mkdir(dirName_model)
return dirName_model
class EarlyStopping:
def __init__(self, logger, patience=7, verbose=False, delta=0):
self.patience = patience
self.verbose = verbose
self.counter = 0
self.best_score = None
self.early_stop = False
self.val_loss_min = np.Inf
self.delta = delta
self.logger = logger
def __call__(self, val_loss, model, path):
score = -val_loss
if self.best_score is None:
self.best_score = score
self.save_checkpoint(val_loss, model, path)
elif score < self.best_score + self.delta:
self.counter += 1
self.logger.info(f'EarlyStopping counter: {self.counter} out of {self.patience}')
if self.counter >= self.patience:
self.early_stop = True
else:
self.best_score = score
self.save_checkpoint(val_loss, model, path)
self.counter = 0
def save_checkpoint(self, val_loss, net, path):
if self.verbose:
print(f'Validation loss decreased ({self.val_loss_min:.6f} --> {val_loss:.6f}). Saving model ...')
torch.save({
'encoder_state_dict': net.encoder.state_dict(),
'decoder_state_dict': net.decoder.state_dict(),
'feature_state_dict': net.feature.state_dict(),
'optimizer_state_dict': net.opt.state_dict(),
}, path)
self.val_loss_min = val_loss
def fillZero(df, type=0):
zeroIdx = df[df.isin([0]).any(axis=1)].index
if type is 0:
nextIdx = zeroIdx + pd.offsets.Hour(1)
prevIdx = zeroIdx + pd.offsets.Hour(-1)
df.loc[zeroIdx] = (df.loc[nextIdx].values + df.loc[prevIdx].values) / 2
else:
nextIdx = zeroIdx + pd.DateOffset(1)
prevIdx = zeroIdx + pd.DateOffset(-1)
df.loc[zeroIdx] = (df.loc[nextIdx].values + df.loc[prevIdx].values) / 2
return df
data_dict = dict()
class Dataset_ISO(Dataset):
def __init__(self, logger, flag='train', past_T=24*7, future_T=24,
data_path='ISONE', target=("load",), scalerX=None, scalerY=None, debug=False):
self.logger = logger
self.past_T = past_T
self.future_T = future_T
# initial
assert flag in ['train', 'test', 'val']
type_map = {'train': 0, 'val': 1, 'test': 2}
self.set_type = type_map[flag]
self.target = target
self.data_path = data_path
if self.set_type != 0 and (scalerX is None or scalerY is None):
self.logger.error("Vali or Test without scaler")
os.exit(-1)
self.scalerX = scalerX
self.scalerY = scalerY
self.debug = debug
self.__read_data__()
def __read_data__(self):
global data_dict
if self.data_path not in data_dict:
self.scalerX = StandardScaler()
self.scalerY = StandardScaler()
df_raw = pd.read_csv(os.path.join("data",
self.data_path+".csv"))
date_rng = pd.date_range(start='1/1/2015 00:00:00', end='12/31/2019 23:00:00', freq='H')
df_raw = df_raw.iloc[-365 * 24 * 5 - 25:-1, -df_raw.shape[1] + 3:].set_index(date_rng)
df_raw = create_datetype_column(df_raw)
self.logger.info(f"features: {df_raw.columns.values}.")
# df_raw.to_csv('isodata.csv',index=False)
else:
df_raw = data_dict[self.data_path]
data_len = df_raw.shape[0]
if self.debug:
border1s = [0, 10 * 24, 20*24]
border2s = [10 * 24, 20 * 24, 30*24]
else:
border1s = [0, 365*24*3+24, data_len-365*24]
border2s = [365*24*3+192+24, 365*24*4+24, data_len]
border1 = border1s[self.set_type]
border2 = border2s[self.set_type]
proc_dat = df_raw[border1:border2].values
mask = np.ones(proc_dat.shape[1], dtype=bool) # 82 true
dat_cols = list(df_raw.columns) # get all column name
for col_name in self.target:
mask[dat_cols.index(col_name)] = False
feats = proc_dat[:, mask]
targs = proc_dat[:, ~mask]
self.numFeatures = df_raw.columns.get_loc("load")
if self.data_path not in data_dict:
self.scalerX.fit(feats[:, :self.numFeatures])
self.scalerY.fit(targs)
data_dict[self.data_path] = df_raw
feats_scaled = self.scalerX.transform(feats[:, :self.numFeatures])
targs_scaled = self.scalerY.transform(targs)
feats_scaled_combine = np.concatenate([feats_scaled, feats[:, self.numFeatures:]], axis=1)
self.feats = feats_scaled_combine
self.targs = targs_scaled
self.targs_ori = targs
self.featureSize = self.feats.shape[1]
def __getitem__(self, index):
s_begin = index
s_end = s_begin + self.past_T
r_begin = s_end
r_end = r_begin + self.future_T
feats = self.feats[s_begin:s_end]
target_feats = self.feats[r_begin:r_end]
y_history = self.targs[s_begin:s_end]
y_target = self.targs[r_begin:r_end]
return feats, y_history, y_target, target_feats
def __len__(self):
return len(self.feats) - self.past_T - self.future_T + 1
def inverse_transform(self, data):
return self.scalerY.inverse_transform(data)
class Dataset_Utility(Dataset):
def __init__(self, logger, flag='train', past_T=24*7, future_T=24,
data_path='Utility', target=("load",), scalerX=None, scalerY=None, debug=False):
self.logger = logger
self.past_T = past_T
self.future_T = future_T
# initial
assert flag in ['train', 'test', 'val']
type_map = {'train': 0, 'val': 1, 'test': 2}
self.set_type = type_map[flag]
self.target = target
self.data_path = data_path
if self.set_type != 0 and (scalerX is None or scalerY is None):
self.logger.error("Vali or Test without scaler")
os.exit(-1)
self.scalerX = scalerX
self.scalerY = scalerY
self.debug = debug
self.__read_data__()
def __read_data__(self):
global data_dict
if self.data_path not in data_dict:
self.scalerX = StandardScaler()
self.scalerY = StandardScaler()
df_temp = pd.read_fwf(os.path.join("data", "input.txt"), header=None).iloc[:-1,:]
df_load = pd.read_fwf(os.path.join("data", "output.txt"), header=None).iloc[:-1,:]
df_temp[0]=pd.to_datetime(df_temp[0], format='%m/%d/%y')
df_load[0]=pd.to_datetime(df_load[0], format='%m/%d/%y')
df_temp=df_temp.set_index(0)
df_load = df_load.set_index(0)
df_temp = df_temp.loc['1987-1-1':'1991-12-31'].values.reshape(-1)
df_load = df_load.loc['1987-1-1':'1991-12-31'].values.reshape(-1)
date_rng = pd.date_range(start='1/1/1987 00:00:00', end='12/31/1991 23:00:00', freq='H')
df_raw = pd.DataFrame({'T':df_temp,'load':df_load}, index=date_rng)
df_raw = fillZero(df_raw)
df_raw = create_datetype_column(df_raw)
self.logger.info(f"features: {df_raw.columns.values}.")
else:
df_raw = data_dict[self.data_path]
data_len = df_raw.shape[0]
if self.debug:
border1s = [0, 10 * 24, 20*24]
border2s = [10 * 24, 20 * 24, 30*24]
else:
border1s = [0, 365*24*3+24, data_len-365*24]
border2s = [365*24*3+192+24, 365*24*4+24, data_len]
border1 = border1s[self.set_type]
border2 = border2s[self.set_type]
proc_dat = df_raw[border1:border2].values
# proc_dat = dat
mask = np.ones(proc_dat.shape[1], dtype=bool) # 82 true
dat_cols = list(df_raw.columns) # get all column name
for col_name in self.target:
mask[dat_cols.index(col_name)] = False
feats = proc_dat[:, mask]
targs = proc_dat[:, ~mask]
self.numFeatures = df_raw.columns.get_loc("load")
if self.data_path not in data_dict:
self.scalerX.fit(feats[:, :self.numFeatures])
self.scalerY.fit(targs)
data_dict[self.data_path] = df_raw
feats_scaled = self.scalerX.transform(feats[:, :self.numFeatures])
targs_scaled = self.scalerY.transform(targs)
feats_scaled_combine = np.concatenate([feats_scaled, feats[:, self.numFeatures:]], axis=1)
self.feats = feats_scaled_combine
self.targs = targs_scaled
self.targs_ori = targs
self.featureSize = self.feats.shape[1]
def __getitem__(self, index):
s_begin = index
s_end = s_begin + self.past_T
r_begin = s_end
r_end = r_begin + self.future_T
feats = self.feats[s_begin:s_end]
target_feats = self.feats[r_begin:r_end]
y_history = self.targs[s_begin:s_end]
y_target = self.targs[r_begin:r_end]
return feats, y_history, y_target, target_feats
def __len__(self):
return len(self.feats) - self.past_T - self.future_T + 1
def inverse_transform(self, data):
return self.scalerY.inverse_transform(data)
class testSampler(Sampler):
def __init__(self, length):
self.length = length
def __iter__(self):
return iter(range(0,self.length,24))
def __len__(self) -> int:
return len(range(0,self.length,24))
class Dataset_Update(Dataset):
def __init__(self, logger, feats, targs, targs_pred, targs_ori, past_T, future_T):
self.logger = logger
self.feats = feats
self.targs = targs
self.targs_pred = targs_pred
self.targs_ori = targs_ori
self.err = self.targs - self.targs_pred
self.past_T = past_T
self.future_T = future_T
def __getitem__(self, index):
s_begin = index
s_end = s_begin + self.past_T
r_begin = s_end
r_end = r_begin + self.future_T
feats = self.feats[s_begin:s_end]
target_feats = self.feats[r_begin:r_end]
err = self.err[s_begin:s_end]
y_history = self.targs[s_begin:s_end]
y_target = self.targs[r_begin:r_end]
err_target = self.err[r_begin:r_end]
targs_pred = self.targs_pred[r_begin:r_end]
y_target_ori = self.targs_ori[r_begin:r_end]
return feats, err, y_target, target_feats, err_target, targs_pred, y_target_ori, y_history
def __len__(self):
return len(self.feats) - self.past_T - self.future_T + 1