深度学习花书 笔记2-二项逻辑回归模型推导与编写代码（Binomial logistic regression model

1. 原理推导

2. 逻辑回归二分类代码

首先随机创建一些数据

'''
Created on Apr 5, 2020
Author: ***
'''
import os, sys
import numpy as np
import matplotlib
import matplotlib.pyplot as plt

n = 200 #数据的个数
fw = open('data.txt', 'w')
classes = [0, 1]

x1 = np.random.uniform(1, 10, 100)
y1 = np.random.uniform(1, 10, 100)
for i in range(100):
    fw.write("%f\t%f\t%d\n" % (x1[i], y1[i], classes[0]))

x2 = np.random.uniform(11, 20, 100)
y2 = np.random.uniform(11, 20, 100)
for i in range(100):
    fw.write("%f\t%f\t%d\n" % (x2[i], y2[i], classes[1]))

fw.close()
figure = plt.figure()
ax = figure.add_subplot(111)
ax.scatter(x1, y1, marker='o',s=80, c='green')
ax.scatter(x2, y2, marker='^',s=80, c='red')
plt.show()

数据显示，如下图， 共两类数据

下面是具体的编程实现逻辑回归：

'''
Created on Apr 5, 2020
Author: ****
'''
import os, sys
import numpy as np
import matplotlib
import matplotlib.pyplot as plt

def LoadDataSet(fileName, delim='\t'):
    fr = open(fileName)
    stringArr = [line.strip().split(delim) for line in fr.readlines()]
    datArr = [list(map(float, line)) for line in stringArr]
    return np.mat(datArr)

def BuildMat(data):
    data_build = np.zeros_like(data)
    data_build[:, 2] = 1   # 需要给输入向量加一个偏置项，对应权值参数中的b，即 w*x +b
    data_build[:, 0:2] = data[:, :-1]
    return data_build

def InitParameter(data):
    labels = data[:, -1]           # 获取数据类别
    rows, cols = data.shape
    weights = np.ones((cols, 1))   # 初始化权重向量[w1, w2, b]
    data_build = BuildMat(data)
    return labels, weights, data_build

def LogisticRegression(F):
    sigmoid = 1.0 / (1.0 + np.exp(-F))  # sigmoid 函数
    return sigmoid

def Train(data, labels, weights, epoch, learning_rate):
    weights_list = []
    for i in range(epoch):
        wx = data * np.mat(weights)  # w * x
        output = LogisticRegression(wx)
        loss = labels - output
        weights = weights + learning_rate * data.T * loss  # 根据推导出来的似然函数的导数进行编写
        weights_list.append(weights)
    return weights

def Classifier(test_data, weights):
    F = np.sum(test_data * weights)  # wx + b
    prob = LogisticRegression(F)
    if prob > 0.5:
        cur_class = 1
    else:
        cur_class = 0
    return cur_class

def DisplayData(data):
    figure = plt.figure()
    ax = figure.add_subplot(111)
    for i in range(len(data)):
        if data[i, 2] == 0:
            ax.scatter(data[i, 0], data[i, 1], marker='o', s=80, c='green')
        if data[i, 2] == 1:
            ax.scatter(data[i, 0], data[i, 1], marker='^', s=80, c='red')

if __name__ == '__main__':
    # 1. 读取数据
    data = LoadDataSet('data.txt')
    # 2. 将数据显示出来
    DisplayData(data)
    # 3. 初始化参数
    labels, weights, data_build = InitParameter(data)
    # 4. 训练
    weights = Train(data_build, labels, weights, epoch=1000, learning_rate=0.001)
    # 5. 画出超平面
    X = np.linspace(-5, 25, 301)
    weights = np.array(weights)
    Y = -(weights[2] + X * weights[1]) / weights[0]
    plt.plot(X, Y)
    print(weights)
    # 6. 分类测试
    testdata = BuildMat(np.mat([11,   11, 1.0]))
    classes = Classifier(testdata, weights)
    print(classes)
    if classes == 1:
        plt.plot(testdata[0, 0], testdata[0, 1], '^', color='black', markersize=10)
    else:
        plt.plot(testdata[0, 0], testdata[0, 1], 'o', color='black', markersize=10)

    testdata = BuildMat(np.mat([5,  11,  0.0]))
    classes = Classifier(testdata, weights)
    print(classes)
    if classes == 1:
        plt.plot(testdata[0, 0], testdata[0, 1], '^', color='black', markersize=10)
    else:
        plt.plot(testdata[0, 0], testdata[0, 1], 'o', color='black', markersize=10)
    plt.show()

测试结果如下， 如果分类为第一类，则标记为黑色的三角形，如果为第0类，则标记为黑色的圆形。