自定义数据集 使用scikit-learn中svm的包实现svm分类

import numpy as np
import matplotlib.pyplot as plt

# 定义数据
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]])

# 将两个特征单独提取
x1_data = np.concatenate((class1_points[:, 0], class2_points[:, 0]))
x2_data = np.concatenate((class1_points[:, 1], class2_points[:, 1]))
y = np.concatenate((np.ones(class1_points.shape[0]), -np.ones(class2_points.shape[0])))

# 4. 参数初始化
w1 = 0.1
w2 = 0.1
b = 0
learning_rate = 0.05

# 5、损失函数只是为了展示公式，实际会直接求解导数，代码无用
l_data = x1_data.size

# 显示子图
fig, (ax1, ax2) = plt.subplots(2, 1)

step_list = np.array([])  # 初始化为空数组
loss_values = np.array([])  # 初始化为空数组

# 6、开始迭代
num_iterations = 1000
for n in range(1, num_iterations + 1):
    z = w1 * x1_data + w2 * x2_data + b
    yz = y * z
    loss = 1 - yz
    loss[loss < 0] = 0  # 使用Hinge Loss（合页损失）函数
    hinge_loss = np.mean(loss)
    loss_values = np.append(loss_values, hinge_loss)  # 将损失值添加到数组中
    step_list = np.append(step_list, n)  # 将步数添加到数组中

    # 计算梯度
    gradient_w1 = 0
    gradient_w2 = 0
    gradient_b = 0
    for i in range(len(y)):
        if loss[i] > 0:
            gradient_w1 += -y[i] * x1_data[i]
            gradient_w2 += -y[i] * x2_data[i]
            gradient_b += -y[i]

    gradient_w1 /= len(y)
    gradient_w2 /= len(y)
    gradient_b /= len(y)

    # 更新参数
    w1 -= learning_rate * gradient_w1
    w2 -= learning_rate * gradient_w2
    b -= learning_rate * gradient_b

    # 显示频率设置
    frequence_display = 50
    if n % frequence_display == 0 or n == 1:

        if np.abs(w2) < 1e-5:
            continue
        x1_min, x1_max = 0, 6
        x2_min, x2_max = -(w1 * x1_min + b) / w2, -(w1 * x1_max + b) / w2
        ax1.clear()
        ax1.scatter(x1_data[:len(class1_points)], x2_data[:len(class1_points)], c='red', label='Class 1')
        ax1.scatter(x1_data[len(class1_points):], x2_data[len(class1_points):], c='blue', label='Class 2')
        ax1.plot((x1_min, x1_max), (x2_min, x2_max), 'r-')
        ax1.set_title(f"SVM: w1={round(w1.item(), 3)}, w2={round(w2.item(), 3)}, b={round(b.item(), 3)}")

        # 绘制损失图
        ax2.clear()
        ax2.plot(step_list, loss_values, 'g-')
        ax2.set_xlabel("Step")
        ax2.set_ylabel("Loss")
        # 显示图形
        plt.pause(1)
        # # 清空Jupyter输出
        # display.clear_output(wait=True)

plt.show()