导入必要的库：

import numpy as np
import torch
import torch.nn as nn
import torch.optim as optim
from sklearn.metrics import precision_score, recall_score, f1_score

准备数据：

class1_points = np.array([[1.9, 1.2],
                          [1.5, 2.1],
                          [1.9, 0.5],
                          [1.5, 0.9],
                          [0.9, 1.2],
                          [1.1, 1.7],
                          [1.4, 1.1]])

class2_points = np.array([[3.2, 3.2],
                          [3.7, 2.9],
                          [3.2, 2.6],
                          [1.7, 3.3],
                          [3.4, 2.6],
                          [4.1, 2.3],
                          [3.0, 2.9]])
# 提取两类特征，输入特征维度为2
x1_data = np.concatenate((class1_points[:, 0], class2_points[:, 0]), axis=0)
x2_data = np.concatenate((class1_points[:, 1], class2_points[:, 1]), axis=0)
label = np.concatenate((np.zeros(len(class1_points)), np.ones(len(class2_points))), axis=0)

# 将数据转换为 PyTorch 张量
X = torch.tensor(np.column_stack((x1_data, x2_data)), dtype=torch.float32)
y = torch.tensor(label, dtype=torch.float32).view(-1, 1)

定义模型：

# 定义逻辑回归模型
class LogisticRegression(nn.Module):
    def __init__(self):
        super(LogisticRegression, self).__init__()
        self.linear = nn.Linear(2, 1)  # 输入特征维度为 2，输出为 1

    def forward(self, x):
        return torch.sigmoid(self.linear(x))

损失函数和优化器：

# 初始化模型、损失函数和优化器
model = LogisticRegression()
criterion = nn.BCELoss()  # 二分类交叉熵损失
optimizer = optim.SGD(model.parameters(), lr=0.01)

训练模型及保存：

# 训练模型
epochs = 5000
for epoch in range(epochs):
    model.train()
    optimizer.zero_grad()
    outputs = model(X)
    loss = criterion(outputs, y)
    loss.backward()
    optimizer.step()

    if (epoch + 1) % 100 == 0:
        print(f'Epoch [{epoch + 1}/{epochs}], Loss: {loss.item():.4f}')
# 保存模型
torch.save(model.state_dict(), 'model3.pth')
print("模型已保存")

模型加载及预测：

# 加载模型
loaded_model = LogisticRegression()
loaded_model.load_state_dict(torch.load('model3.pth',
                                        map_location=torch.device('cpu'),
                                        weights_only=True))
loaded_model.eval()

# 进行预测
with torch.no_grad():
    predictions = loaded_model(X)
    predicted_labels = (predictions > 0.5).float()

# 展示预测结果和实际结果
print("实际结果：", y.numpy().flatten())
print("预测结果：", predicted_labels.numpy().flatten())

计算精确度和召回率及F1分数：

# 计算精确度、召回率和 F1 分数
precision = precision_score(y.numpy().flatten(), predicted_labels.numpy().flatten())
recall = recall_score(y.numpy().flatten(), predicted_labels.numpy().flatten())
f1 = f1_score(y.numpy().flatten(), predicted_labels.numpy().flatten())

print(f"精确度（Precision）: {precision:.4f}")
print(f"召回率（Recall）: {recall:.4f}")
print(f"F1 分数: {f1:.4f}")

结果展示：