bp神经网络tensorflow代码 股价

以下是使用 TensorFlow 实现 BP 神经网络预测股价的示例代码：

import tensorflow as tf
import pandas as pd
import numpy as np
import matplotlib.pyplot as plt

# 读取数据
data = pd.read_csv('stock_price.csv')

# 数据预处理
data = data[::-1]  # 倒序
data = data.reset_index(drop=True)  # 重设索引
data['Date'] = pd.to_datetime(data['Date'])  # 转换为时间格式
data = data.set_index('Date')  # 将日期列作为索引
data = data.drop(['Open', 'High', 'Low', 'Adj Close', 'Volume'], axis=1)  # 只保留收盘价列
data = data.dropna()  # 删除缺失值

# 划分训练集和测试集
train_size = int(len(data) * 0.8)
train_data = data.iloc[:train_size]
test_data = data.iloc[train_size:]

# 归一化
train_min = np.min(train_data)
train_max = np.max(train_data)
train_data = (train_data - train_min) / (train_max - train_min)
test_data = (test_data - train_min) / (train_max - train_min)

# 构建数据集
def create_dataset(data, window_size):
    X, y = [], []
    for i in range(len(data) - window_size):
        X.append(data.iloc[i:i+window_size])
        y.append(data.iloc[i+window_size])
    return np.array(X), np.array(y)

window_size = 30
train_X, train_y = create_dataset(train_data, window_size)
test_X, test_y = create_dataset(test_data, window_size)

# 构建模型
model = tf.keras.Sequential([
    tf.keras.layers.Dense(128, input_shape=(window_size,), activation='relu'),
    tf.keras.layers.Dense(64, activation='relu'),
    tf.keras.layers.Dense(1)
])

model.compile(optimizer='adam', loss='mse')

# 训练模型
model.fit(train_X, train_y, epochs=100)

# 预测
train_predict = model.predict(train_X)
test_predict = model.predict(test_X)

# 反归一化
train_predict = train_predict * (train_max - train_min) + train_min
train_y = train_y * (train_max - train_min) + train_min
test_predict = test_predict * (train_max - train_min) + train_min
test_y = test_y * (train_max - train_min) + train_min

# 可视化
plt.plot(train_y, label='train')
plt.plot(train_predict, label='train_predict')
plt.plot(test_y, label='test')
plt.plot(test_predict, label='test_predict')
plt.legend()
plt.show()

此代码将股价的收盘价作为输入，采用 BP 神经网络进行预测。输入数据经过归一化后，使用窗口大小为 30 的滑动窗口构建数据集，模型采用 3 层全连接层，输出预测值。训练过程中使用均方误差作为损失函数，使用 Adam 优化器进行优化。最后将预测结果反归一化后进行可视化。