-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathdata_management.py
239 lines (169 loc) · 8.68 KB
/
data_management.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
import pandas as pd
import datetime
import numpy as np
import os
import sys
# dtype
DTYPE = 'float32'
# reduce if your RAM is not sufficient
DATASET_USAGE = 1.0
# train split after resampling
TRAIN_SPLIT = 0.9
TRAIN_SPLIT_EFF = 0
# resample intervall in minutes
RESAMPLE_INTERVALL = 60
DATE_COLUMS = ['time_p', 'time_unp', 'time_fin']
# RELEVANT_COLUMNS = ['c_battery_size_max', 'c_kombi_current_remaining_range_electric', 'soc_p', 'soc_unp', 'delta_km', 'c_temperature', 'delta_kwh']
# default norm ranges
NORM_RANGE = {'c_battery_size_max': (0, 100000), 'c_kombi_current_remaining_range_electric': (0, 500), 'soc_p': (0, 100), 'soc_unp': (0, 100), 'delta_km': (0, 500), 'c_temperature': (-20, 40), 'delta_kwh': (0, 100)}
# dat to the prepared dataset
PATH_PREP = 'data/prep.csv'
DATASET = None
def readPrepData ():
return pd.read_csv(PATH_PREP, sep=';', parse_dates=DATE_COLUMS)
def getModelPath(name, cell_size, label_type, target_length, step_size):
'''
returns the path for the model containing the model specific parameters
'''
return os.path.join(
'models',
name + '_' + str(cell_size) + '__label_' + label_type + '__target_' + str(target_length) + '__step_' + str(step_size))
# loads the DATASET in PATH_PREP
def loadData():
global DATASET
global TRAIN_SPLIT
print('Reading data ...')
# read the data file
DATASET = readPrepData()
# set time_p index
DATASET.set_index('time_p', inplace=True, drop=False)
# reduce dataset if neccessary
if DATASET_USAGE < 1.0:
DATASET = DATASET[:int(len(DATASET.index) * DATASET_USAGE)]
def filterPLZ(dataset, PLZ):
dataset['PLZ'] = dataset['PLZ'].astype(str)
return dataset[dataset['PLZ'].str[0:len(PLZ)] == PLZ]
# normalized a given number in a given range to [0, 1]
def normalizeNumber(x, bounds):
space = bounds[1] - bounds[0]
return (x - bounds[0]) / (space)
def denormalizeNumber(x, bounds):
space = bounds[1] - bounds[0]
return bounds[0] + x * space
def normalizeDatetime(x):
return ( x.hour * 60 + x.minute ) / 24 / 60 # normalize daytime on minutely bases
def getNormWeekOfYear(x):
return x.weekofyear / 52
def getNormDayOfWeek(x):
return x.dayofweek / 7
def normalizeData(dataset, label_type, intervall=RESAMPLE_INTERVALL):
intervall = datetime.timedelta(minutes=intervall)
global NORM_RANGE
global TRAIN_SPLIT_EFF
print('Resample and normalize data ...')
# init nor data with time
norm_data = pd.DataFrame(dataset['time_p'].apply(normalizeDatetime).resample(intervall, label='right', closed='right').min())
# add week of year
norm_data['week_of_year'] = dataset['time_p'].apply(getNormWeekOfYear).resample(intervall, label='right', closed='right').mean()
norm_data['day_of_week'] = dataset['time_p'].apply(getNormDayOfWeek).resample(intervall, label='right', closed='right').mean()
# normalize and resample temperature [-20, +40]
norm_data['c_temperature'] = dataset['c_temperature'].apply(normalizeNumber, args=[NORM_RANGE['c_temperature']]).resample(intervall, label='right', closed='right').mean()
# normalize times in week
# for col in DATE_COLUMS:
# dataset[col] = dataset[col].apply(normalizeDatetime)
if label_type == 'kwh' or label_type == 'minutes_charged':
# normalize and resample batterysize [0, 100000]
norm_data['c_battery_size_max'] = dataset['c_battery_size_max'].resample(intervall, label='right', closed='right').sum()
NORM_RANGE['c_battery_size_max'] = (0, norm_data['c_battery_size_max'].max())
norm_data['c_battery_size_max'] = norm_data['c_battery_size_max'].apply(normalizeNumber, args=[NORM_RANGE['c_battery_size_max']])
# normalize and resample SOC [0, 100]
norm_data['soc_p'] = dataset['soc_p'].apply(normalizeNumber, args=[NORM_RANGE['soc_p']]).resample(intervall, label='right', closed='right').mean()
norm_data['soc_unp'] = dataset['soc_unp'].apply(normalizeNumber, args=[NORM_RANGE['soc_unp']]).resample(intervall, label='right', closed='right').mean()
if label_type == 'kwh':
# normalize and resample delta kwh
norm_data['delta_kwh'] = dataset['delta_kwh'].resample(intervall, label='right', closed='right').sum()
NORM_RANGE['delta_kwh'] = (0, norm_data['delta_kwh'].max())
norm_data['delta_kwh'] = norm_data['delta_kwh'].apply(normalizeNumber, args=[NORM_RANGE['delta_kwh']])
# normalize and resample delta kwh
if label_type == 'count':
norm_data['count'] = dataset['time_p'].resample(intervall, label='right', closed='right').count()
NORM_RANGE['count'] = (0, norm_data['count'].max())
norm_data['count'] = norm_data['count'].apply(normalizeNumber, args=[NORM_RANGE['count']])
# normalize minutes charged
if label_type == 'minutes_charged':
norm_data['minutes_charged'] = dataset['minutes_charged'].resample(intervall, label='right', closed='right').sum()
NORM_RANGE['minutes_charged'] = (0, norm_data['minutes_charged'].max())
norm_data['minutes_charged'] = norm_data['minutes_charged'].apply(normalizeNumber, args=[NORM_RANGE['minutes_charged']])
# fill all nans
norm_data.fillna(0, inplace=True)
# adjust train split
TRAIN_SPLIT_EFF = int(len(norm_data.index) * TRAIN_SPLIT)
return norm_data
# sum of loaded kwh plugged after current time
def getKWHLabel(df):
return df['delta_kwh'].sum()
# counts the the events in the target time frame
def getCountLabel(df):
return df['count'].sum()
# counts the the events in the target time frame
def getMinutesChargedLabel(df):
return df['minutes_charged'].sum()
# returns the label dependent on the selected label type
def getLabel(df, labelType, current_time, target, step):
label = []
relevant_time = current_time + datetime.timedelta(seconds=1)
for t in range(target):
if labelType == 'kwh':
label.append(getKWHLabel(df[relevant_time + t * step: relevant_time + (t+1) * step]))
if labelType == 'count':
label.append(getCountLabel(df[relevant_time + t * step: relevant_time + (t+1) * step]))
if labelType == 'minutes_charged':
label.append(getMinutesChargedLabel(df[relevant_time + t * step: relevant_time + (t+1) * step]))
return label
def getTFDataset(dataset, history, target, lable_type, step=0 ):
data = []
labels = []
start_date = dataset.index[0] + datetime.timedelta(minutes=history)
end_date = dataset.index[-1] - datetime.timedelta(minutes=target*step)
if step == 0:
step = datetime.timedelta(minutes=RESAMPLE_INTERVALL)
step = datetime.timedelta(minutes=step)
count = 0
print('Labeling data ...')
while start_date < end_date:
data.append(np.array(dataset[start_date-datetime.timedelta(minutes=history):start_date], dtype=DTYPE))
labels.append(np.array(getLabel(dataset[start_date:start_date+step*target], lable_type, start_date, target, step), dtype=DTYPE))
count+=1
sys.stdout.write("\r{0} - {1} {2} time steps labeled -> Label: {3}".format(start_date, count, step, labels[-1][0]))
sys.stdout.flush()
start_date += step
print('\n')
return np.array(data), np.array(labels)
# returns a training dataset based on history in minutes, target in minutes and label type
def getTrainDataset(history, target_time, label_type, step=0):
global DATASET
return getTFDataset(DATASET[:TRAIN_SPLIT_EFF], history, target_time, label_type, step=step)
# returns a test dataset based on history in minutes, target in minutes and label type
def getValDataset(history, target_time, label_type, step=0):
global DATASET
return getTFDataset(DATASET[TRAIN_SPLIT_EFF:], history, target_time, label_type, step=step)
def getTestData(timestamp, history, target, label_type, step, PLZ=None):
global DATASET
loadData()
if PLZ != None:
DATASET = filterPLZ(DATASET, PLZ)
data = DATASET[timestamp-datetime.timedelta(minutes=history):timestamp].copy()
DATASET = normalizeData(DATASET, label_type, step)
norm_data = np.array([np.array(DATASET[timestamp-datetime.timedelta(minutes=history):timestamp].copy(), dtype=DTYPE)])
label = np.array(getLabel(DATASET[timestamp:timestamp+datetime.timedelta(minutes=target*step)], label_type, timestamp, target, datetime.timedelta(minutes=step)), dtype=DTYPE)
return data, norm_data, label
def getEvaluationData(target, label_type, step):
global DATASET
loadData()
DATASET = normalizeData(DATASET, label_type, step)[TRAIN_SPLIT_EFF:]
return DATASET
# inits the data management
def init(label_type, resample_intervall=RESAMPLE_INTERVALL):
global DATASET
loadData()
DATASET = normalizeData(DATASET, label_type, resample_intervall)