python实现lstm多步预测

LSTM（Long Short-Term Memory）是一种常用的循环神经网络，可以用于时间序列数据的建模和预测。多步预测是指预测多个时间步长的未来值。下面是Python实现LSTM多步预测的示例代码。

首先，导入所需的库：

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.preprocessing import MinMaxScaler
from keras.models import Sequential
from keras.layers import Dense, LSTM

接着，读取并处理数据。假设我们有一个时间序列数据集，包含10年每月销售额的信息。

# 读取数据
data = pd.read_csv('sales.csv', header=None)

# 数据归一化
scaler = MinMaxScaler(feature_range=(0, 1))
data = scaler.fit_transform(data)

然后，将数据集划分为训练集和测试集。

# 划分训练集和测试集
train_size = int(len(data) * 0.67)
test_size = len(data) - train_size
train, test = data[0:train_size,:], data[train_size:len(data),:]

接下来，将训练集和测试集转换为可以用于LSTM模型训练的输入和输出。

# 将数据转换为可以用于LSTM模型训练的输入和输出
def create_dataset(dataset, look_back=1, look_forward=1):
    dataX, dataY = [], []
    for i in range(len(dataset)-look_back-look_forward+1):
        a = dataset[i:(i+look_back), 0]
        dataX.append(a)
        dataY.append(dataset[(i+look_back):(i+look_back+look_forward), 0])
    return np.array(dataX), np.array(dataY)

# 设置时间步长
look_back = 12
look_forward = 3

# 创建训练数据集
trainX, trainY = create_dataset(train, look_back, look_forward)

# 创建测试数据集
testX, testY = create_dataset(test, look_back, look_forward)

然后，定义LSTM模型并进行训练。

# 定义LSTM模型
model = Sequential()
model.add(LSTM(4, input_shape=(1, look_back)))
model.add(Dense(look_forward))
model.compile(loss='mean_squared_error', optimizer='adam')
model.fit(trainX.reshape((trainX.shape[0], 1, trainX.shape[1])), trainY, epochs=100, batch_size=1, verbose=2)

最后，使用训练好的模型进行多步预测。

# 使用训练好的模型进行多步预测
trainPredict = model.predict(trainX.reshape((trainX.shape[0], 1, trainX.shape[1])))
testPredict = model.predict(testX.reshape((testX.shape[0], 1, testX.shape[1])))

# 反归一化
trainPredict = scaler.inverse_transform(trainPredict)
trainY = scaler.inverse_transform(trainY)
testPredict = scaler.inverse_transform(testPredict)
testY = scaler.inverse_transform(testY)

完整代码如下：

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
from sklearn.preprocessing import MinMaxScaler
from keras.models import Sequential
from keras.layers import Dense, LSTM

# 读取数据
data = pd.read_csv('sales.csv', header=None)

# 数据归一化
scaler = MinMaxScaler(feature_range=(0, 1))
data = scaler.fit_transform(data)

# 划分训练集和测试集
train_size = int(len(data) * 0.67)
test_size = len(data) - train_size
train, test = data[0:train_size,:], data[train_size:len(data),:]

# 将数据转换为可以用于LSTM模型训练的输入和输出
def create_dataset(dataset, look_back=1, look_forward=1):
    dataX, dataY = [], []
    for i in range(len(dataset)-look_back-look_forward+1):
        a = dataset[i:(i+look_back), 0]
        dataX.append(a)
        dataY.append(dataset[(i+look_back):(i+look_back+look_forward), 0])
    return np.array(dataX), np.array(dataY)

# 设置时间步长
look_back = 12
look_forward = 3

# 创建训练数据集
trainX, trainY = create_dataset(train, look_back, look_forward)

# 创建测试数据集
testX, testY = create_dataset(test, look_back, look_forward)

# 定义LSTM模型
model = Sequential()
model.add(LSTM(4, input_shape=(1, look_back)))
model.add(Dense(look_forward))
model.compile(loss='mean_squared_error', optimizer='adam')
model.fit(trainX.reshape((trainX.shape[0], 1, trainX.shape[1])), trainY, epochs=100, batch_size=1, verbose=2)

# 使用训练好的模型进行多步预测
trainPredict = model.predict(trainX.reshape((trainX.shape[0], 1, trainX.shape[1])))
testPredict = model.predict(testX.reshape((testX.shape[0], 1, testX.shape[1])))

# 反归一化
trainPredict = scaler.inverse_transform(trainPredict)
trainY = scaler.inverse_transform(trainY)
testPredict = scaler.inverse_transform(testPredict)
testY = scaler.inverse_transform(testY)