-
Notifications
You must be signed in to change notification settings - Fork 0
/
LSTM_7MAIO.py
163 lines (109 loc) · 4.3 KB
/
LSTM_7MAIO.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
import numpy as np
import pandas as pd
import os
import matplotlib.pyplot as plt
import tensorflow as tf
from tensorflow.keras import Sequential, Model
from tensorflow.keras.layers import Layer, Dense, Dropout, LSTM
from sklearn.utils import shuffle
from sklearn.preprocessing import MinMaxScaler
from tensorflow.compat.v1.keras.layers import CuDNNLSTM
df = pd.read_csv('Dataset_7MAIO.csv', delimiter = ',',
encoding = 'ISO-8859-1')
# Ordenar por mês, dia e hora.
df.sort_values(['Month (number)', 'Day of month', 'Hour'],
ascending = [True, True, True], inplace = True)
# Separação por ruas.
df_1 = df[df['road_num'] == 1]
df_2 = df[df['road_num'] == 2]
df_3 = df[df['road_num'] == 3]
df_4 = df[df['road_num'] == 4]
SPLIT_INDEX = len(df_1) - 8*24
df_1.drop('road_num', axis = 1, inplace = True)
# Vamos tratar da rua 1.
#dataset = df_1_train.dropna(subset=["speed_diff"])
#dataset=dataset.reset_index(drop=True)
df_1=df_1.reset_index(drop=True)
df_1_train = df_1[:SPLIT_INDEX]
df_1_test = df_1[SPLIT_INDEX:]
'''training_set = df_1.iloc[:,4:5].values # Só contem valores do speed_diff
sc = MinMaxScaler(feature_range=(0,1))
training_set_scaled = sc.fit_transform(training_set)'''
'''
We will create a training set such that for every 7 days (7*24 hours) we will provide the next 24 hours
speed_diff as output. In other words, input for our RNN would be 7 days temperature data and
the output would be 1 day forecast of speed_diff
'''
x_train = []
y_train = []
n_future = 24 # next 4 days temperature forecast
n_past = 24*7 # Past 30 days
label = df_1_train['speed_diff']
for i in range(0,len(df_1_train)-n_past-n_future+1):
x_train.append(df_1_train.iloc[i : i + n_past])
y_train.append(label.iloc[i + n_past : i + n_past + n_future ])
for i in range(len(x_train)):
x_train[i]=np.array(x_train[i])
for i in range(len(y_train)):
y_train[i]=np.array(y_train[i])
for i in range(len(x_train)):
x_train[i] = np.reshape(x_train[i], (x_train[0].shape[0] , x_train[0].shape[1]) )
for i in range(len(y_train)):
y_train[i] = np.reshape(y_train[i], (y_train[0].shape[0]))
x_train=np.array(x_train)
y_train=np.array(y_train)
scalers=[]
for i in range(17):
sc = MinMaxScaler(feature_range=(0,1))
x_train[:,i] = sc.fit_transform(x_train[:,i])
scalers.append(sc)
print('x_train',len(x_train),len(x_train[0]))
sc1 = MinMaxScaler(feature_range=(0,1))
y_train = sc1.fit_transform(y_train)
regressor = Sequential()
regressor.add(LSTM(units=24*7, return_sequences=True, input_shape = (168,17) ) )
regressor.add(Dropout(0.2))
regressor.add(LSTM(24*7 , return_sequences=True))
regressor.add(Dropout(0.2))
regressor.add(LSTM(24*7, return_sequences=True))
regressor.add(Dropout(0.2))
regressor.add(LSTM(24*7))
regressor.add(Dropout(0.2))
regressor.add(Dense(24,activation='sigmoid'))
regressor.compile(optimizer='adam', loss='mean_squared_error',metrics=['acc'])
regressor.fit(x_train, y_train, epochs=1 )
##############################
# read test dataset
testdataset = df_1_test
#get only the temperature column
testdataset = testdataset.iloc[:24*7]
x_test = []
label1 = testdataset['speed_diff']
x_test.append(testdataset.iloc[0 : n_past])
for i in range(len(x_test)):
x_test[i]=np.array(x_test[i])
for i in range(len(x_test)):
x_test[i] = np.reshape(x_test[i], (x_test[0].shape[0] , x_test[0].shape[1]) )
print(x_test)
x_test=np.array(x_test)
real_temperature = df_1_test
real_temperature = real_temperature.iloc[24*7:]
print(real_temperature)
y_real = []
label2 = real_temperature['speed_diff']
y_real.append(label2.iloc[:n_future ])
for i in range(len(y_real)):
y_real[i]=np.array(y_real[i])
for i in range(len(y_real)):
y_real[i] = np.reshape(y_real[i], (y_real[0].shape[0]))
for i in range(17):
x_test[:,i] = scalers[i].fit_transform(x_test[:,i])
print(y_real)
print('############################')
#testing = sc.transform(testdataset)
#testing = np.reshape(x_test,(testing.shape[1],testing.shape[0]))
predicted_temperature = regressor.predict(x_test)
predicted_temperature = sc1.inverse_transform(predicted_temperature)
predicted_temperature = np.reshape(predicted_temperature,(predicted_temperature.shape[1],predicted_temperature.shape[0]))
for i in range(len(predicted_temperature)):
print('Valor real na hora: ',i ,y_real[0][i],'Valor previsto: ',predicted_temperature[i][0])