Python从0到100（八十八）：LSTM网络详细介绍及实战指南

# 简单RNN单元实现
class SimpleRNNCell:
    def __init__(self, input_size, hidden_size):
        self.Wxh = np.random.randn(hidden_size, input_size)*0.01  # 输入到隐藏层权重
        self.Whh = np.random.randn(hidden_size, hidden_size)*0.01  # 隐藏层到隐藏层权重
        self.bh = np.zeros((hidden_size, 1))  # 隐藏层偏置
        
    def forward(self, x, h_prev):
        h_next = np.tanh(np.dot(self.Wxh, x) + np.dot(self.Whh, h_prev) + self.bh)
        return h_next

import yfinance as yf

data = yf.download('AAPL', start='2010-01-01', end='2023-12-31')
features = data[['Open', 'High', 'Low', 'Close', 'Volume']].values

from sklearn.preprocessing import MinMaxScaler

scaler = MinMaxScaler(feature_range=(0, 1))
scaled_data = scaler.fit_transform(features)

# 创建时间序列样本
def create_dataset(data, look_back=60):
    X, y = [], []
    for i in range(len(data)-look_back-1):
        X.append(data[i:(i+look_back)])
        y.append(data[i+look_back, 3])  # 预测Close价格
    return np.array(X), np.array(y)

X, y = create_dataset(scaled_data)

from tensorflow.keras.models import Sequential
from tensorflow.keras.layers import LSTM, Dense, Dropout

model = Sequential([
    LSTM(128, return_sequences=True, input_shape=(X.shape[1], X.shape[2])),
    Dropout(0.3),
    LSTM(64, return_sequences=False),
    Dropout(0.3),
    Dense(32, activation='relu'),
    Dense(1)
])

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

history = model.fit(
    X_train, y_train,
    epochs=100,
    batch_size=32,
    validation_split=0.2,
    callbacks=[
        EarlyStopping(monitor='val_loss', patience=10),
        ReduceLROnPlateau(monitor='val_loss', factor=0.1, patience=5)
    ]
)

from tensorflow.keras.layers import Bidirectional

Bidirectional(LSTM(64, return_sequences=True))

from tensorflow.keras.layers import Embedding

# 文本生成模型
model = Sequential([
    Embedding(vocab_size, 256),
    LSTM(1024, return_sequences=True),
    LSTM(512),
    Dense(vocab_size, activation='softmax')
])

# 完整的时间序列预测代码
import numpy as np
import tensorflow as tf
from sklearn.preprocessing import MinMaxScaler

# 数据准备
def load_data(file_path, look_back=60):
    # 实现数据加载和预处理
    pass  

# 构建LSTM模型
def build_lstm_model(input_shape):
    model = tf.keras.Sequential([
        tf.keras.layers.LSTM(128, return_sequences=True, input_shape=input_shape),
        tf.keras.layers.Dropout(0.3),
        tf.keras.layers.LSTM(64),
        tf.keras.layers.Dropout(0.3),
        tf.keras.layers.Dense(32, activation='relu'),
        tf.keras.layers.Dense(1)
    ])
    model.compile(optimizer='adam', loss='mse')
    return model

# 模型训练与评估
if __name__ == "__main__":
    X_train, y_train, X_test, y_test = load_data('stock_data.csv')
    model = build_lstm_model((X_train.shape[1], X_train.shape[2]))
    history = model.fit(X_train, y_train, epochs=100, validation_split=0.2)
    predictions = model.predict(X_test)
    # 实现评估指标计算和可视化