简单对照统计学习上的感知机理论，基于numpy实现感知机学习模型

1代码

import pandas as pd
import numpy as np
from matplotlib import pyplot as plt


def fit(x,y,max_iter=100, tol=1e-3, eta0=0.1):
    y = y.reshape(-1, 1)
    if x.shape[0]!=y.shape[0] or y.shape[1]!=1:
        raise ValueError

    iter_count = 0
    w = np.ones(shape=(x.shape[1],1))
    b = 0
    loss = (np.dot(x, w) + b) * y
    loss_function_sum = -1 * np.sum(loss[np.where(loss < 0)])
    error_lst = list(np.where(loss < 0)[0])

    while (iter_count<max_iter and loss_function_sum>tol and len(error_lst)>0):
        loss = (np.dot(x, w) + b) * y
        loss_function_sum = -1*np.sum(loss[np.where(loss<0)])
        iter_count += 1
        error_lst = list(np.where(loss < 0)[0])

        if len(error_lst)>0:
            error_x = x[error_lst[0]]
            error_y = y[error_lst[0]]

            w_step = (eta0 * error_x * error_y).reshape(-1, 1)
            b_step = (eta0*error_y)[0]
            w += w_step
            b += b_step
            b = round(b,6)
        else:
            continue
    return w,b

if __name__ == '__main__':
    filePath = r'D:\statisticsLearn\perception\data'
    data = pd.read_csv(filePath + r'\sample.csv')
    data['y'] = data['y'].apply(lambda x:2*x-1)

    x = data[['x1','x2']].values
    y = data['y'].values

    # 训练数据和测试数据
    x_data_train = x[:80, :]
    x_data_test = x[80:, :]
    y_data_train = y[:80]
    y_data_test = y[80:]

    # 正例和反例
    positive_x1 = [x[i, 0] for i in range(100) if y[i] == 1]
    positive_x2 = [x[i, 1] for i in range(100) if y[i] == 1]
    negetive_x1 = [x[i, 0] for i in range(100) if y[i] == -1]
    negetive_x2 = [x[i, 1] for i in range(100) if y[i] == -1]

    coef = fit(x,y,max_iter=1000,eta0=1)

    plt.scatter(positive_x1, positive_x2, c='red')
    plt.scatter(negetive_x1, negetive_x2, c='blue')
    # 画出超平面（在本例中即是一条直线）
    line_x = np.arange(-4, 4)
    line_y = line_x * (coef[0][0] / coef[0][1]) - coef[1]
    plt.plot(line_x, line_y)
    plt.show()

2与sklearn比较

Figure_1.png