-
Notifications
You must be signed in to change notification settings - Fork 6
/
Copy pathkerasLSTMStockNto1.py
194 lines (179 loc) · 6.32 KB
/
kerasLSTMStockNto1.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
# -*- coding: utf-8 -*-
"""
Created on Wed Jan 1 21:17:06 2020
@author: 64054
"""
import os
import sys
#数据预处理以及绘制图形需要的模块
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
##
from sklearn.preprocessing import MinMaxScaler
from keras.models import Sequential
from keras.layers import Dense, LSTM, BatchNormalization
###############################################################################
def get_dataset():
#从csv读取数据
dataset = pd.read_csv('./000001.csv', encoding = 'gb18030')
#print(dataset)
#数据集的维度
print(dataset.shape)
#将数据价转换为整数
dataset['开盘价'] = dataset['开盘价'].astype(int)
dataset['收盘价'] = dataset['收盘价'].astype(int)
dataset['成交量'] = dataset['成交量'].astype(int)
# 列名选取行，剔除不需要开头的行
dataframeDataset = dataset.loc[1:, ['开盘价', '收盘价', '成交量']]
print(dataframeDataset.shape)
# 反转数据
dataframeDatasetReverse = dataframeDataset.reindex(index = dataframeDataset.index[::-1])
# 历史日期在前
ndarrayDataset = dataframeDatasetReverse.values
#print(ndarrayDataset)
#画图大盘数据图
#plt.plot(ndarrayDataset[:,0], ndarrayDataset[:,1])
#plt.show()
return ndarrayDataset
def get_lstm_dataset(dataset, need_num, total_dataset):
lstm_dataset = []
for i in range(need_num, total_dataset):
lstm_dataset.append(dataset[i-need_num:i])
lstm_dataset = np.array(lstm_dataset)
#print(xTrain)
#print(lstm_dataset.shape)
return lstm_dataset
###############################################################################
#需要之前5次的股票数据来预测下一次的数据，
need_num = 5
#训练数据的大小
training_num = 240
#测试数据的大小
test_num = 1
#迭代训练10次
epoch = 10
#每次取数据数量
batch_size = 10
###############################################################################
#数据处理
ndarrayDataset = get_dataset()
###############################################################################
# 构建训练集
x_train = ndarrayDataset[0:training_num]
#print(x_train)
y_train = []
for i in range(1, training_num+1):
y_train.append(ndarrayDataset[i,0])
#print(y_train)
# 转换dt数据
x_train, y_train = np.array(x_train), np.array(y_train)
print(x_train.shape)
print(y_train.shape)
###############################################################################
#数据归一化,fit_transform处理后就变成list
sc_X = MinMaxScaler(feature_range=(0, 1))
x_train = sc_X.fit_transform(x_train)
x_train = np.array(x_train)
print(x_train.shape)
#print(x_train)
#print(x_train.shape)
# outcome scaling:
sc_Y = MinMaxScaler(feature_range=(0, 1))
y_train = sc_Y.fit_transform(y_train.reshape(-1,1))
y_train = np.array(y_train)
print(y_train.shape)
###############################################################################
#构建LSTM需要的数据
xTrain = []
for i in range(need_num, training_num):
xTrain.append(x_train[i-need_num:i])
xTrain = np.array(xTrain)
#print(xTrain)
print(xTrain.shape)
x_train = xTrain
yTrain = []
for i in range(need_num, training_num):
yTrain.append(y_train[i])
yTrain = np.array(yTrain)
#print(yTrain)
print(yTrain.shape)
y_train = yTrain
#因为LSTM要求输入的数据格式为三维的，[training_number, time_steps, 1]，因此对数据进行相应转化
#x_train = np.reshape(x_train, (x_train.shape[0], x_train.shape[1], 1))
###############################################################################
#构建网络，使用的是序贯模型
model = Sequential()
#return_sequences=True返回的是全部输出，LSTM做第一层时，需要指定输入shape
model.add(LSTM(units=128, return_sequences=True,
input_shape=[x_train.shape[1],
x_train.shape[2]]))
model.add(BatchNormalization())
model.add(LSTM(units=128))
model.add(BatchNormalization())
model.add(Dense(units=1))
#进行配置
model.compile(optimizer='adam',
loss='mean_squared_error',
metrics=['accuracy'])
###############################################################################
#训练网络
model.fit(x=x_train, y=y_train, epochs=epoch, batch_size=batch_size)
###############################################################################
#进行测试数据的处理
#将训练集的数据向后移动，增加上需要预测的数据
x_test = ndarrayDataset[test_num:training_num+test_num]
y_test = []
for i in range(test_num+1, training_num+test_num+1):
y_test.append(ndarrayDataset[i,0])
#print(y_train)
# 转换dt数据
x_test, y_test = np.array(x_test), np.array(y_test)
print(x_test.shape)
print(y_test.shape)
###############################################################################
#测试数据的归一化
sc_X_T = MinMaxScaler(feature_range=(0, 1))
x_test = sc_X_T.fit_transform(x_test)
x_test = np.array(x_test)
#print(x_test)
print(x_test.shape)
sc_Y_T = MinMaxScaler(feature_range=(0, 1))
y_test_transform = sc_Y_T.fit_transform(y_test.reshape(-1,1))
y_test_transform = np.array(y_test_transform)
#print(y_test_transform)
print(y_test_transform.shape)
#获取对应的开盘价列
###############################################################################
#构建LSTM需要的数据
xTest = []
for i in range(need_num, training_num):
xTest.append(x_test[i-need_num:i])
xTest = np.array(xTest)
#print(xTrain)
print(xTest.shape)
x_test = xTest
yTest = []
for i in range(need_num, training_num):
yTest.append(y_test[i])
yTest = np.array(yTest)
#print(yTrain)
print(yTest.shape)
y_test = yTest
###############################################################################
#进行预测
y_predictes = model.predict(x=x_test)
print(y_predictes)
print(y_predictes.shape)
#使用 sc.inverse_transform()将归一化的数据转换回原始的数据，以便我们在图上进行查看
y_predictes = sc_Y_T.inverse_transform(X=y_predictes)
print(y_predictes)
###############################################################################
#绘制数据图表，红色是真实数据，蓝色是预测数据
plt.plot(y_test, color='red', label='Real Stock Price')
plt.plot(y_predictes, color='blue', label='Predicted Stock Price')
plt.title(label='ShangHai Stock Price Prediction')
plt.xlabel(xlabel='Time')
plt.ylabel(ylabel='ShangHai Stock Price')
plt.legend()
plt.show()