时间序列数据分析：特征选择、LSTM模型训练及模型保存

加载时间序列数据集

import pandas as pd

data = pd.read_csv('time_series_data.csv')

对时间序列数据集选择最佳特征

from sklearn.feature_selection import SelectKBest
from sklearn.feature_selection import f_regression

# 选择最佳的5个特征
selector = SelectKBest(f_regression, k=5)
X = data.iloc[:, :-1]
y = data.iloc[:, -1]
X_new = selector.fit_transform(X, y)

使用 PyTorch 库对时间序列数据集的最佳特征进行 LSTM 模型训练

import torch
import torch.nn as nn

# 定义LSTM模型
class LSTM(nn.Module):
    def __init__(self, input_size, hidden_size, num_layers, output_size):
        super(LSTM, self).__init__()
        self.hidden_size = hidden_size
        self.num_layers = num_layers
        self.lstm = nn.LSTM(input_size, hidden_size, num_layers, batch_first=True)
        self.fc = nn.Linear(hidden_size, output_size)

    def forward(self, x):
        h0 = torch.zeros(self.num_layers, x.size(0), self.hidden_size).to(device)
        c0 = torch.zeros(self.num_layers, x.size(0), self.hidden_size).to(device)

        out, _ = self.lstm(x, (h0, c0))

        out = self.fc(out[:, -1, :])

        return out

# 定义模型参数
input_size = X_new.shape[1]
hidden_size = 10
num_layers = 2
output_size = 1

# 初始化模型和优化器
model = LSTM(input_size, hidden_size, num_layers, output_size).to(device)
criterion = nn.MSELoss()
optimizer = torch.optim.Adam(model.parameters(), lr=0.001)

# 训练模型
num_epochs = 1000
for epoch in range(num_epochs):
    inputs = torch.from_numpy(X_new).float().to(device)
    targets = torch.from_numpy(y.values).float().to(device)

    optimizer.zero_grad()

    outputs = model(inputs)

    loss = criterion(outputs, targets)
    loss.backward()

    optimizer.step()

    if (epoch+1) % 100 == 0:
        print(f'Epoch [{epoch+1}/{num_epochs}], Loss: {loss.item():.4f}')

保存模型参数数据文件

PATH = 'lstm_model.pth'
torch.save(model.state_dict(), PATH)

注意：

• 请确保将 'time_series_data.csv' 替换为您的实际数据文件路径。
• 该代码示例假设您已经安装了必要的库，例如 pandas、scikit-learn 和 PyTorch。
• 您可能需要根据您的具体需求调整代码中的参数，例如特征选择数量、模型结构等。