pathon实现感知机分类乳腺癌数据集，并用混淆矩阵显示结果

 先上代码：

from matplotlib import pyplot as plt
from sklearn.datasets import load_breast_cancer
from sklearn.metrics import confusion_matrix
from sklearn.model_selection import train_test_split
import numpy as np
import seaborn as sns

cancers = load_breast_cancer()


def sign(w, x, b):
    return np.sign(np.dot(w, x) + b)


def train1(feature, label, w, b, num, learning_rate):
    for j in range(num):
        count = feature.shape[0]
        for i in range(feature.shape[0]):
            print(i)
            if sign(w, feature[i], b) * label[i] <= 0:  # <0定义为分类错误，>0定义为分类正确
                print("分类错误！误分类点为:", feature[i])
                w = w + learning_rate * label[i] * feature[i]
                b = b + learning_rate * label[i]
            else:
                count -= 1
    w = np.around(w, 2) # 设置小数点保留位数
    b = np.around(b, 2)
    print("最终权重 w:", w, "最终偏置 b:", b, "分类错误个数", count)
    return w, b


def heatmap(cancers):
    x = cancers.data  # 获取特征值
    y = cancers.target  # 获取标签
    learning_rate = 0.1  # 学习率
    x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.2)
    # 之前是0和1的时候会出问题，在更新权重那边
    for i in range(len(y_test)):
        if y_test[i] == 0:
            y_test[i] = -1
    for k in range(len(y_train)):
        if y_train[k] == 0:
            y_train[k] = -1
    print(y_train)
    weight = np.zeros(x.shape[1])
    bias = 0

    w, b = train1(x_train, y_train, weight, bias, 100, learning_rate)
    y_testhat = np.sign(np.dot(x_test, w) + b)
    result = confusion_matrix(y_test, y_testhat)
    sns.set(font_scale=1.5)
    # 热力图主要展示的是二维数据的数据关系
    sns.heatmap(data=result, square=True)
    plt.xlabel('Predict label', fontsize=18)
    plt.ylabel('True label', fontsize=18)
    plt.show()  # 显示热度图


heatmap(cancers)

代码解释：

1.在train1函数中原本运行终止结果应该是没有分类错误样本，但之前运行了很久也没有成功跑出来，就设置了一定的迭代次数。

2.在获取标签后将0换成了-1，否则分类错误时的样本在更新时一直是加0（即label[i] * feature[i]）

运行结果：