pytorch-实现LSTM实现多变量输入多步预测

 话不多说，我们直接上代码，在专栏《时间序列预测项目实战》第一篇文章里，已经讲到使用LSTM实现单变量输入单步预测的问题，那么如果在一般的情况下是可以满足需求的。

但在实际的项目需求中或者学术研究中，不乏会出现要使用多变量输入的情况进行预测，且这种预测不仅仅局限于单步预测，也会是多步预测，因此，本篇内容主要是用来实现多变量的输入，且用于多步的预测输出。

☆后续会更新完基础模型后，会继续对最新的前沿模型进行讲解与代码实战，帮助大家一起进行时序预测这个领域的研究，希望大家多多支持，作者会加快更新。

import matplotlib.pyplot as plt
import numpy as np
import pandas as pd
import torch
import torch.nn as nn
import tushare as ts
from sklearn.preprocessing import StandardScaler, MinMaxScaler
from torch.utils.data import TensorDataset
from tqdm import tqdm

class Config():
    data_path = '../data/wind_dataset.csv'
    timestep = 20  # 时间步长，就是利用多少时间窗口
    batch_size = 32  # 批次大小
    feature_size = 8  # 每个步长对应的特征数量，这里只使用1维，每天的风速
    hidden_size = 256  # 隐层大小
    output_size = 2  # 由于是多输出任务，最终输出层大小为2，预测未来2天风速
    num_layers = 2  # lstm的层数
    epochs = 10 # 迭代轮数
    best_loss = 0 # 记录损失
    learning_rate = 0.0003 # 学习率
    model_name = 'lstm' # 模型名称
    save_path = './{}.pth'.format(model_name) # 最优模型保存路径

config = Config()

# 1.加载时间序列数据
df = pd.read_csv(config.data_path, index_col = 0)


# 2.将数据进行标准化
scaler = MinMaxScaler()
scaler_model = MinMaxScaler()
data = scaler_model.fit_transform(np.array(df))
scaler.fit_transform(np.array(df['WIND']).reshape(-1, 1))

# 形成训练数据，例如12345789 12345-67 23456-78
def split_data(data, timestep, feature_size, output_size):
    dataX = []  # 保存X
    dataY = []  # 保存Y

    # 将整个窗口的数据保存到X中，将未来一天保存到Y中
    for index in range(len(data) - timestep - 1):
        dataX.append(data[index: index + timestep])
        dataY.append(data[index + timestep: index + timestep + output_size][:, 0].tolist())
        

    dataX = np.array(dataX)
    dataY = np.array(dataY)
    
    # 获取训练集大小
    train_size = int(np.round(0.8 * dataX.shape[0]))

    # 划分训练集、测试集
    x_train = dataX[: train_size, :].reshape(-1, timestep, feature_size)
    y_train = dataY[: train_size].reshape(-1, output_size)
    
    x_test = dataX[train_size:, :].reshape(-1, timestep, feature_size)
    y_test = dataY[train_size:].reshape(-1, output_size)

    return [x_train, y_train, x_test, y_test]

# 3.获取训练数据   x_train: 170000,30,1   y_train:170000,7,1
x_train, y_train, x_test, y_test = split_data(data, config.timestep, config.feature_size, config.output_size)

# 4.将数据转为tensor
x_train_tensor = torch.from_numpy(x_train).to(torch.float32)
y_train_tensor = torch.from_numpy(y_train).to(torch.float32)
x_test_tensor = torch.from_numpy(x_test).to(torch.float32)
y_test_tensor = torch.from_numpy(y_test).to(torch.float32)

# 5.形成训练数据集
train_data = TensorDataset(x_train_tensor, y_train_tensor)
test_data = TensorDataset(x_test_tensor, y_test_tensor)

# 6.将数据加载成迭代器
train_loader = torch.utils.data.DataLoader(train_data,
                                           config.batch_size,
                                           False)

test_loader = torch.utils.data.DataLoader(test_data,
                                          config.batch_size,
                                          False)

# 7.定义LSTM网络
class LSTM(nn.Module):
    def __init__(self, feature_size, hidden_size, num_layers, output_size):
        super(LSTM, self).__init__()
        self.hidden_size = hidden_size  # 隐层大小
        self.num_layers = num_layers  # lstm层数
        # feature_size为特征维度，就是每个时间点对应的特征数量，这里为1
        self.lstm = nn.LSTM(feature_size, hidden_size, num_layers, batch_first=True)
        self.fc = nn.Linear(hidden_size, output_size)

    def forward(self, x, hidden=None):
        batch_size = x.shape[0] # 获取批次大小
        
        # 初始化隐层状态
        if hidden is None:
            h_0 = x.data.new(self.num_layers, batch_size, self.hidden_size).fill_(0).float()
            c_0 = x.data.new(self.num_layers, batch_size, self.hidden_size).fill_(0).float()
        else:
            h_0, c_0 = hidden
            
        # LSTM运算
        output, (h_0, c_0) = self.lstm(x, (h_0, c_0))
        
        # 全连接层
        output = self.fc(output)  # 形状为batch_size * timestep, 1
        
        # 我们只需要返回最后一个时间片的数据即可
        return output[:, -1, :]

model = LSTM(config.feature_size, config.hidden_size, config.num_layers, config.output_size)  # 定义LSTM网络
loss_function = nn.MSELoss()  # 定义损失函数
optimizer = torch.optim.AdamW(model.parameters(), lr=config.learning_rate)  # 定义优化器

# 8.模型训练
for epoch in range(config.epochs):
    model.train()
    running_loss = 0
    train_bar = tqdm(train_loader)  # 形成进度条
    for data in train_bar:
        x_train, y_train = data  # 解包迭代器中的X和Y
        optimizer.zero_grad()
        y_train_pred = model(x_train)
        loss = loss_function(y_train_pred, y_train)
        loss.backward()
        optimizer.step()

        running_loss += loss.item()
        train_bar.desc = "train epoch[{}/{}] loss:{:.3f}".format(epoch + 1,
                                                                 config.epochs,
                                                                 loss)

    # 模型验证
    model.eval()
    test_loss = 0
    with torch.no_grad():
        test_bar = tqdm(test_loader)
        for data in test_bar:
            x_test, y_test = data
            y_test_pred = model(x_test)
            test_loss = loss_function(y_test_pred, y_test)

    if test_loss < config.best_loss:
        config.best_loss = test_loss
        torch.save(model.state_dict(), save_path)

print('Finished Training')

# 9.绘制结果
plot_size = 200
plt.figure(figsize=(12, 8))
plt.plot(scaler.inverse_transform((model(x_train_tensor).detach().numpy()[: plot_size, 0]).reshape(-1, 1)), "b")
plt.plot(scaler.inverse_transform(y_train_tensor.detach().numpy().reshape(-1, 1)[: plot_size]), "r")
plt.legend()
plt.show()

y_test_pred = model(x_test_tensor)
plt.figure(figsize=(12, 8))
plt.plot(scaler.inverse_transform(y_test_pred.detach().numpy()[: plot_size, 0].reshape(-1, 1)), "b")
plt.plot(scaler.inverse_transform(y_test_tensor.detach().numpy().reshape(-1, 1)[: plot_size]), "r")
plt.legend()
plt.show()