-
Notifications
You must be signed in to change notification settings - Fork 0
/
Long short-term memory for public transport latness.py
288 lines (197 loc) · 13.5 KB
/
Long short-term memory for public transport latness.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
#!/usr/bin/env python
# coding: utf-8
# In[47]:
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import datetime
import time
import multiprocessing as mp
from itertools import product
from keras.models import Sequential,Model
from keras.layers import Dense, Input, Concatenate, Lambda,LSTM,Dropout,Layer
from keras.optimizers import RMSprop
from sys import getsizeof
import warnings
from functions import *
# # Data import
# In[2]:
startTime = 7*3600
endTime = 10*3600
dt = 30
Data = get_time_data('data/data2/Times/fahrzeiten_soll_ist_20200105_20200111.csv')
Data = Data[Data['datum_nach'] == Data['datum_nach'][1]]
#delete depot run
Data = Data[~Data['fw_lang'].str.contains('DEP|Einfahrt|Ausfahrt')]
#Data=Data[(Data['soll_ab_nach']>startTime ) & (Data['soll_ab_nach']<endTime)] # only during peak hour (6am to 9 am)
#,'datum_nach','fahrzeug','soll_ab_nach'])
Studied_line = 32
connected = []
Stops = get_stops_data('data/data2/haltestelle.csv')
Stops_pos = get_stops_data('data/data2/haltepunkt.csv')
Line = get_line_data('data/data1/linie.csv')
Stops['Line'] = ""
Line['Connected'] = 0
for stop in Stops['halt_diva']:
if any(Data.loc[Data['halt_diva_nach'] == stop ,'linie'].unique() == Studied_line):
for l in [Data.loc[Data['halt_diva_nach'] == stop ,'linie'].unique()][0]:
Line.loc[Line['Linienname'] == l,'Connected'] = 1
Line.loc[Studied_line,'Connected'] = 2
Stops_pos = Stops_pos[Stops_pos['halt_punkt_ist_aktiv']]
Stops_pos = Stops_pos.groupby(by = 'halt_id').mean()
Stops_pos = Stops_pos.merge(Stops[{'halt_id', 'halt_diva'}],left_on='halt_id',right_on='halt_id')
if all(Data.columns != 'Connected'):
Data = Data.merge(Line,left_on='linie',right_on='Linienname')
Data = Data.merge(Stops_pos[{'halt_diva','GPS_Latitude',
'GPS_Longitude'}],left_on='halt_diva_nach',right_on='halt_diva')
# ## features selection but here we do not select features but stops
#
# In[22]:
get_ipython().run_line_magic('matplotlib', 'notebook')
time_frame = 20*60 # we look at Xmin data frame
Bus_id = Data.loc[(Data['linie'] == Studied_line),'fahrzeug'].unique()[0]
Data_bus = Data[(Data['fahrzeug'] == Bus_id)]
Data_bus['Stop'] = (Data_bus['richtung']-1) * (max(Data_bus['seq_nach'])+1 - Data_bus['seq_nach']) - (Data_bus['richtung']-2) * Data_bus['seq_nach']
Data_bus = Data_bus.sort_values(by='ist_ab_von')
plt.plot(Data_bus['ist_ab_von']/60,Data_bus['Stop'],color = 'r')
plt.plot(Data_bus['soll_ab_von']/60,Data_bus['Stop'],color = 'b')
y_real = Data[Data['fahrzeug'] == Bus_id]# all the delays for one precise bus (ID = bus number)
#Features.groupby(['Line']).mean()
select = np.array(['Delta_stops','Delta_trip','Tot_lat','Is_connected','IsBus','IsTram','IsOther','Time','Dist','Lat','ID'])
#y_value = Data[]
# ## Convert data to feed them in the neural network
# In this section all the data are converted to have bus stops delays at any point in time. The idea is too look every dt, if there is a change we update, if not we keep the same value. The first iteration the last known position is kept. We extract the following data for each day ;
#
# + Increase in lateness due to boarding
# + Increase in lateness due to travel time
# + Total lateness
# + If the line is connected to our line
# + the type of the line
# + the Time
# + the distance to the studied vehicle
#
# Note that the data are shape is made x: time and y stops and then transposed, the process was much more faster ( more than 10x) because panda prefer to work with a lot of row than lots of columns
#
# # NOTES TO MYSELF,
# to win time on the change of type, transform T-B etc to float. same with is connected, should we had a 2 for is line ?
#
# dt or each new studied bus arrival ? - does LSTM like having not regular time step ?
# In[23]:
idx = pd.IndexSlice
ncoeur = 8
Data = Data.sort_values(by =['Linienname','halt_diva_nach'])
t1 = time.time()
BusPos = pd.DataFrame(columns = range(startTime,endTime,dt),index = ['Lat','Long'])
#FrameT = pd.DataFrame(index = range(startTime,endTime,dt),columns = range(Data['datum_nach'].unique().shape[0])) & Data['datum_nach'] == d
#Colnames = [np.repeat(range(startTime,endTime,dt),11),np.resize(select,11*int((endTime-startTime)/dt))]
# def dataSelection(Data):
for i in [1]:
print(type(Data))
Colnames = range(startTime,endTime,dt)
RowNames = [np.repeat(10000*Data[{'Linienname','halt_diva_nach'}].drop_duplicates()['Linienname'] + Data[{'Linienname','halt_diva_nach'}].drop_duplicates()['halt_diva_nach'],11),
np.resize(select,11*len(Data[{'Linienname','halt_diva_nach'}].drop_duplicates()))]
FrameT = pd.DataFrame(index = RowNames,columns = Colnames)
BusPos = pd.DataFrame(columns = range(startTime,endTime,dt),index = ['Lat','Long'])
for t in range(startTime,endTime,dt):
if t == startTime:
# the first t is lower so we have the last known position
sel = 10000*Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),{'Linienname','halt_diva_nach'}].drop_duplicates()['Linienname']+Data.loc[(Data['ist_ab_nach']>t-3600 ) & (Data['ist_ab_nach']<t+dt),{'Linienname','halt_diva_nach'}].drop_duplicates()['halt_diva_nach']
FrameT.loc[idx[sel,:],t] = pd.DataFrame(
{'Delta_stops' :(Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'ist_ab_nach']-Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'ist_an_nach1']-Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'soll_ab_nach']+Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'soll_an_nach']),
'Delta_trip' :Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'ist_an_nach1']-Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'ist_ab_von']-Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'soll_an_nach']+Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'soll_ab_von'],
'Tot_lat' :Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'ist_ab_nach']-Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'soll_ab_nach'],
'Is_connected' :Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'Connected'].values,
'IsBus' : ((Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'VSYS'].values == 'B' )|(
Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'VSYS'].values == 'TB')).astype('float'),
'IsTram' :(Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'linie'] == 'T').astype('float'),
'IsOther' :((Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'VSYS'].values == 'B' ) & (
Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'linie'].values == 'T') & (
Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'VSYS'].values == 'TB')).astype('float'),
'Time' :Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'ist_ab_nach'].values,
'Dist' :Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'GPS_Longitude'].values,
'Lat' :Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'GPS_Latitude'].values,
'ID' : 10000*Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'Linienname']+Data.loc[(Data['ist_ab_nach']>t- 3600 ) & (Data['ist_ab_nach']<t+dt),'halt_diva_nach']}
).drop_duplicates(subset = 'ID',keep = 'last').values.ravel()
if all(FrameT.loc[idx[sel,'ID'],t].values != sel.values):
warnings.warn('ID from Data is not ID from FrameT',DeprecationWarning)
BusPos[t] = Data.loc[(Data['ist_ab_nach']>t -3600 ) & (Data['ist_ab_nach']<t+dt) & (Data['fahrzeug']==Bus_id),['GPS_Latitude','GPS_Longitude']].drop_duplicates(keep = 'last').T.values
else:
FrameT[t]=FrameT[t-dt]
# Bus_pos = Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt) & (Data['fahrzeug'] == Bus_id) ,{'GPS_Longitude','GPS_Latitude'}].
sel = 10000*Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),{'Linienname','halt_diva_nach'}].drop_duplicates()['Linienname']+Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),{'Linienname','halt_diva_nach'}].drop_duplicates()['halt_diva_nach']
FrameT.loc[idx[sel,:],t] = pd.DataFrame(
{'Delta_stops' :(Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'ist_ab_nach']-Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'ist_an_nach1']-Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'soll_ab_nach']+Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'soll_an_nach']),
'Delta_trip' :Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'ist_an_nach1']-Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'ist_ab_von']-Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'soll_an_nach']+Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'soll_ab_von'],
'Tot_lat' :Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'ist_ab_nach']-Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'soll_ab_nach'],
'Is_connected' :Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'Connected'].values,
'IsBus' : ((Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'VSYS'].values == 'B' )|(
Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'VSYS'].values == 'TB')).astype('float'),
'IsTram' :(Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'linie'] == 'T').astype('float'),
'IsOther' :((Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'VSYS'].values == 'B' ) & (
Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'linie'].values == 'T') & (
Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'VSYS'].values == 'TB')).astype('float'),
'Time' :Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'ist_ab_nach'].values,
'Dist' :Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'GPS_Longitude'].values,
'Lat' :Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'GPS_Latitude'].values,
'ID' : 10000*Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'Linienname']+Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt),'halt_diva_nach']}
).drop_duplicates(subset = 'ID',keep = 'last').values.ravel()
if all(FrameT.loc[idx[sel,'ID'],t].values != sel.values):
warnings.warn('ID from Data is not ID from FrameT',DeprecationWarning)
if Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt) & (Data['fahrzeug']==Bus_id),['GPS_Latitude','GPS_Longitude']].empty :
BusPos[t] = BusPos[t-dt]
else:
BusPos[t] = Data.loc[(Data['ist_ab_nach']>t ) & (Data['ist_ab_nach']<t+dt) & (Data['fahrzeug']==Bus_id),['GPS_Latitude','GPS_Longitude']].drop_duplicates(keep = 'last').T.values
# compute the distance to the studied bus
FrameT = FrameT.drop('ID', level=1)
FrameT = FrameT.drop('Line', level=1)
FrameT = FrameT.drop('Stop', level=1)
FrameT.loc[idx[:,'Dist'],:] = abs((FrameT.loc[idx[:,'Dist'],:].subtract(BusPos.loc['Long'].values,axis = 1).values)**2 + (FrameT.loc[idx[:,'Lat'],:].subtract(BusPos.loc['Lat'].values,axis = 1).values)**2)
FrameT = FrameT.drop('Lat', level=1)
# return FrameT
# pool = mp.Pool(ncoeur) #product(range(startTime,endTime,dt)
# i = 0
# for day in Data['datum_nach'].unique():
# print(day)
# if i == 0:
# splitData = list([Data.loc[Data['datum_nach'] == day,:]])
# else:
# splitData.append(Data.loc[Data['datum_nach'] == day,:])
# i+=1
# print('Time spend :',time.time()- t1)
# test = pool.map(dataSelection,splitData)
# pool.close()
# pool.join() # 28
print('Time spend :',time.time()- t1)
# In[43]:
FrameT = FrameT.dropna(axis=0)
FrameT= FrameT.transpose()
t1 = time.time()
FrameT = FrameT.astype('float')
# In[44]:
print(getsizeof(Data))
epochs = 20
batch_size = 200
print(Data.shape)
auto = Sequential([
#Merge data for each stops
Dense(4400, activation='relu',input_shape=(8841,)),
Dense(2200, activation='relu'),
#decoder
Dense(4400, activation='relu'),
Dense(8841, activation='sigmoid')
])
auto.compile(loss='mean_squared_error', optimizer = RMSprop())
auto.summary()
t1 = time.time()
autoencoder_train = auto.fit(FrameT,FrameT, batch_size=
batch_size,epochs=epochs,verbose=1)
# In[48]:
loss = autoencoder_train.history['loss']
epochs_plot = range(epochs)
plt.figure(2)
plt.plot(epochs_plot, loss, 'bo', label='Training loss')
plt.title('Loss')
plt.xlabel('Epochs')
plt.ylabel('Loss')
plt.legend()
plt.show()