python实现PLA——感知机算法（学习笔记）

 1 import numpy as np
 2 import matplotlib.pyplot as plt
 3 from sklearn.datasets import load_iris
 4 from sklearn import datasets
 5 
 6 
 7 def sign(x):
 8     if x < 0:
 9         return -1
10     else:
11         return 1
12 
13 
14 def model(w, x, b):
15     y = (-b - w[0] * x) / w[1]  # 将直线转化为y=ax+b格式
16     return y
17 
18 
19 def fit(data):
20     learning_rate = 0.001   # 学习率
21     w = np.array([0.0, 0.0])
22     b = 0.2
23     sumCount = 0
24     while True:
25         count = 0
26         for i in range(data.shape[0]):
27             print(f‘迭代次数：{sumCount},w: {w},b: {b}‘)
28             if data[i][-1] * sign(np.dot(w, data[i][0:2]) + b) <= 0:
29                 w += data[i][0:2] * data[i][-1] * learning_rate  # 根据计算得出的梯度和设定的学习率来调整w和b
30                 b += data[i][-1] * learning_rate
31                 count += 1
32                 sumCount += 1
33         if count == 0:  # 如果本轮迭代没有发现误分类点，则跳出循环
34             break
35     return w, b
36 
37 
38 iris = datasets.load_iris()
39 iris_data = load_iris().data
40 
41 sepal_length = np.array([iris_data[i][0] for i in range(iris_data.shape[0])])
42 sepal_width = np.array([iris_data[i][1] for i in range(iris_data.shape[0])])
43 # sepal_length为x轴，sepal_width为y轴
44 list = []
45 for i in range(50):  # 划分数据，前50个为一组，后50个为一组
46     list.append([sepal_length[i], sepal_width[i], 1])
47     list.append([sepal_length[50 + i], sepal_width[50 + i], -1])
48 data = np.array(list)
49 
50 plt.xlabel(‘sepal_length‘)
51 plt.xlim(4.2, 7.1)
52 plt.ylabel(‘sepal_width‘)
53 plt.ylim(1.5, 5)
54 for i in range(100):
55     if data[i][-1] == 1:
56         plt.scatter(data[i][0], data[i][1], color=‘blue‘, marker=‘*‘)
57     else:
58         plt.scatter(data[i][0], data[i][1], color=‘red‘, marker=‘o‘)
59 
60 w, b = fit(data)
61 x = np.array([i for i in range(len(sepal_length))])
62 plt.plot(x, model(w, x, b))
63 plt.show()