机器学习系列之逻辑回归

逻辑回归，主要用于二分类问题。其基本形式是每个特征乘以一个回归系数，然后所有值相加。

在代入sigmoid函数中，得到一个离散的分类结果。然后在以0.5（或其他值）作为阈值，进行0-1分类。

极大似然估计：利用已知样本数据，去估计相关的参数。将上诉的内容在利用极大似然估计法，进行估计简化，再利用梯度下降法求解。

梯度下降法：

参考：https://www.cnblogs.com/muchen/p/6296957.html#_labelTop

          https://www.cnblogs.com/lc1217/p/6802637.html

具体实现代码如下：
#!/usr/bin/env python
# -*- coding:utf-8 -*-
__author__ = 'Great'
'''
读取整理数据
（逻辑回归函数）
计算sigmord函数
（最大似然估计）
梯度下降法求解（迭代式）
预测
显示分类效果
'''
from math import exp
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd

#读取数据集，处理训练数据及标签
def data_set():
    x_train = []
    x_label = []
    df = open('./test/testSet.txt')
    for line in df.readlines():
        line_df = line.strip().split()
        x_train.append([1.0, float(line_df[0]), float(line_df[1])])
        x_label.append(float(line_df[2]))
    return x_train, x_label

#signord转换
def sigmord(num):

    return 1.0/(1 + np.exp(-num))#math.exp()不能对矩阵操作

#梯度下降法
def gradt(data, label):
    labelmat = np.mat(label).transpose()
    datamat = np.mat(data)
    m, n = np.shape(datamat)

    alpha = 0.001
    maxtry = 500
    weights = np.ones((n, 1))
    for k in range(maxtry):
        h = sigmord(datamat*weights)
        error = (labelmat - h)
        weights = weights + alpha*datamat.transpose()*error
    return weights

'''
def test():
    data, label = data_set()
    result = gradt(data, label)
    print(result)
'''
def plot_fit(weight):
    data, label = data_set()
    data_array = np.array(data)
    xfeat0 = []
    yfeat0 = []
    xfeat1 = []
    yfeat1 = []
    for i in range(np.shape(data_array)[0]):
        if int(label[i]) == 1:
            xfeat1.append(data_array[i,1])
            yfeat1.append(data_array[i,2])
        else:
            xfeat0.append(data_array[i,1])
            yfeat0.append(data_array[i,2])
    fig = plt.figure()
    ax = fig.add_subplot(1,1,1)
    ax.scatter(xfeat0,xfeat0, s = 30, c = 'green', marker = 'o')
    ax.scatter(xfeat1,yfeat1, s = 30, c = 'red')

    x = np.arange(-3.0, 3.0, 0.1)
    y = (-weight[0] - weight[1]*x)/weight[2]
    ax.plot(x, y)
    plt.xlabel('X1')
    plt.ylabel('X2')
    plt.show()

def test1():
    data, label = data_set()
    weight = gradt(data, label)
    wg = weight.getA()#将numpy转为数组
    plot_fit(wg)

if __name__ =='__main__':
    test1()

#还需要更多关于应用的使用和理解......