机器学习LDA线性判别器代码实现

机器学习LDA线性判别器代码实现

西瓜书P60 线性判别器LDA代码实现：

import numpy as np
import matplotlib.pyplot as plt
def load_data(file_name):
    '''
    数据导入函数
    :param file_name: (string)训练数据位置
    :return: feature_data(mat)特征
            lable_data(mat)标签
    '''
    fr = open(file_name)
    feature_data =[];
    lable_data = [];
    for line in fr.readlines():
        curLine = []
        lineArr = line.split('\t')
        for i in range(0,2):
            curLine.append(float(lineArr[i]))
        feature_data.append(curLine)
        if len(lineArr)<3:
            continue;
        tempLine = []
        for i in range(2,3):
            tempLine.append(int(lineArr[i]))
        lable_data.append(tempLine)
    feature_mat = np.array(feature_data,dtype=float)
    label_mat = np.array(lable_data,dtype=int)
    fr.close()
    return feature_mat,label_mat

def LDA(x1,x2):
    '''
    :param x1: 类别1  （num,d）
    :param x2: 类别2   (num,d)
    :return: 投影向量w
    '''
    u1 = np.mean(x1,axis=0)
    u2 = np.mean(x2,axis=0)
    Sw = np.dot((x1 - u1).T,(x1-u1)) + np.dot((x2 - u2).T,(x2-u2))
    Swmat = np.mat(Sw)
    w = np.dot(Swmat.I,(u1-u2))
    return w


def LDA_2(x1,x2):
    u1 = np.mean(x1, axis=0)
    u2 = np.mean(x2, axis=0)

    s1 = 0
    s2 = 0
    for i in range(0,len(x1)):
        s1 = s1 + np.dot((x1[i,:] - u1).T,(x1[i,:] - u1))
    for i in range(0,len(x2)):
        s2 = s2 + np.dot((x2[i,:] - u2).T ,(x2[i,:] - u2))
    Sw = s1+s2
    Sw = np.mat(Sw)
    w = Sw.I *(u1 - u2).T
    return  w

if __name__ == "__main__":
    # 1. 导入数据
    print("------1. load data------")
    feature_data, lable_data = load_data("train_data.txt")

    x1_data =[]
    x2_data = []
    for i in range(0,len(feature_data)):
        if(lable_data[i] == 1):
            x1_data.append(feature_data[i])
        else:
            x2_data.append(feature_data[i])

    x1_data = np.array(x1_data)
    x2_data = np.array(x2_data)
    w = LDA(x1_data,x2_data)
    # w = LDA_2(x1_data, x2_data)
    print(w)

    """使用scatter()绘制散点图"""

    x_values = range(-5, 5)
    #映射适量的xielv
    rate = w[0,1]/w[0,0]
    #对应垂直于这个直线的斜率
    rate2 = - w[0,0]/w[0,1]
    y_values = [rate * x for x in x_values]
    '''
    scatter() 
    x:横坐标 y:纵坐标 s:点的尺寸
    '''
    #y = kx+b 中的b
    b = feature_data[:, 1] - feature_data[:, 0]*rate2
    # b = feature_data[:, 0] - feature_data[:, 1]*rate2
    #计算出焦点x
    x_ = b/(rate-rate2)
    #计算出焦点y
    y_ = x_*rate2+b

    plt.plot(x_values, y_values)
    plt.scatter(feature_data[:,0],feature_data[:,1] , s=10)
    w = np.array(w)
    # aa =np.squeeze()
    plt.scatter(x_,y_ , s=10)

    # plt.scatter(x_values, y_values, c=y_values, cmap=plt.cm.Blues, edgecolors='none', s=10)

    # 设置图表标题并给坐标轴加上标签
    plt.title('LDA', fontsize=24)
    plt.xlabel('feature_1', fontsize=14)
    plt.ylabel('feature_2', fontsize=14)

    # 设置刻度标记的大小
    plt.tick_params(axis='both', which='major', labelsize=14)
    plt.show()

最终效果 

 

 

 数据集：https://download.csdn.net/download/willen_/11110641