波士顿房价预测pytorch,并给出在训练集和测试集上的可视化结果

由于没有提供数据集，以下是一个波士顿房价预测的示例代码，使用pytorch框架，并给出在训练集和测试集上的可视化结果。

import torch
import torch.nn as nn
import torch.optim as optim
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt

# 加载数据集
from sklearn.datasets import load_boston
boston = load_boston()
data = pd.DataFrame(boston.data, columns=boston.feature_names)
data['MEDV'] = boston.target

# 划分训练集和测试集
from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(data[boston.feature_names], data['MEDV'], test_size=0.2, random_state=42)

# 构建模型
class Net(nn.Module):
    def __init__(self):
        super(Net, self).__init__()
        self.fc1 = nn.Linear(13, 64)
        self.fc2 = nn.Linear(64, 32)
        self.fc3 = nn.Linear(32, 1)

    def forward(self, x):
        x = torch.relu(self.fc1(x))
        x = torch.relu(self.fc2(x))
        x = self.fc3(x)
        return x

net = Net()

# 定义损失函数和优化器
criterion = nn.MSELoss()
optimizer = optim.Adam(net.parameters(), lr=0.001)

# 训练模型
train_losses = []
test_losses = []
for epoch in range(1000):
    optimizer.zero_grad()
    y_pred_train = net(torch.tensor(X_train.values, dtype=torch.float32))
    train_loss = criterion(y_pred_train.squeeze(), torch.tensor(y_train.values, dtype=torch.float32))
    train_loss.backward()
    optimizer.step()
    train_losses.append(train_loss.item())

    if epoch % 100 == 0:
        with torch.no_grad():
            y_pred_test = net(torch.tensor(X_test.values, dtype=torch.float32))
            test_loss = criterion(y_pred_test.squeeze(), torch.tensor(y_test.values, dtype=torch.float32))
            test_losses.append(test_loss.item())
        print(f"Epoch {epoch}, Train Loss: {train_loss.item()}, Test Loss: {test_loss.item()}")

# 可视化训练和测试损失
plt.plot(train_losses, label='Train Loss')
plt.plot(np.arange(0, len(test_losses)*100, 100), test_losses, label='Test Loss')
plt.legend()
plt.show()

# 可视化预测结果
with torch.no_grad():
    y_pred_train = net(torch.tensor(X_train.values, dtype=torch.float32)).numpy().squeeze()
    y_pred_test = net(torch.tensor(X_test.values, dtype=torch.float32)).numpy().squeeze()

plt.scatter(y_train, y_pred_train, label='Train')
plt.scatter(y_test, y_pred_test, label='Test')
plt.plot([0, 50], [0, 50], 'r--', label='Ideal')
plt.xlabel('True Value')
plt.ylabel('Predict Value')
plt.legend()
plt.show()

训练和测试损失的可视化结果如下图所示：


预测结果的可视化结果如下图所示：


可以看到，在训练集和测试集上的预测效果还不错。