python实现Logistic回归

本文不涉及逻辑回归的具体原理，只通过python代码实现算法，并且没有用到机器学习库，根据算法流程一步一步实现。

逻辑回归

数据准备

文件中没条数据有两个属性，和一个标签

# 数据准备
def loadData():
    x = []
    y = []
    fr = open('data.txt')
    for i in fr.readlines():
        i = i.strip().split()    # 分隔
        x.append([1.0, float(i[0]), float(i[1])])   # 输入数据，增加一个输入1
        y.append(int(i[2]))  # 标签
    return x, y

sigmoid函数

# sigmoid函数
def sigmoid(x):
    return 1.0 / (1 + np.exp(-x))

权重更新

采用梯度上升算法来更新权重，又可分为批梯度上升算法，一次性使用全部的数据值

# 批梯度下降，更新权重
def GA(x, y):
    x = np.mat(x)             # 转化为numpy矩阵
    y = np.mat(y).transpose() # 转化为numpy矩阵
    m, n = np.shape(x)  
    alpha = 0.001
    epochs = 500
    weights = np.ones((n, 1))
    for k in range(epochs):
        z = x * weights
        h = sigmoid(z)  # 矩阵乘法
        error = (h - y)           # 残差
        weights = weights - alpha * x.transpose() * error
    return weights

也可使用随机梯度上升算法，一次使用一个数据

# 随机梯度下降，更新权重
def SGA(x, y):
    x = np.array(x)
    m, n = np.shape(x)
    alpha = 0.01
    weights = np.ones(n)  # 初始化权重
    for i in range(m):
        h = sigmoid(sum(x[i] * weights))
        error = h - y[i]
        weights = weights - alpha * error * x[i]
    return weights

上面两种方式的学习率都是固定的，也可是变化的

# 改进的随机梯度下降
def SGA1(x, y, numIter=150):
    m, n = np.shape(x)
    x = np.array(x)
    weights = np.ones(n)
    for j in range(numIter):
        dataIndex = list(range(m))
        for i in range(m):
            alpha = 4/(1.0+j+i)+0.0001   
            randIndex = int(np.random.uniform(0, len(dataIndex)))
            h = sigmoid(sum(x[randIndex]*weights))
            error = h - y[randIndex]
            weights = weights - alpha * error * x[randIndex]
            del(dataIndex[randIndex])
    return weights

分类

输入数据完成分类任务

# 分类
def classify(x, weights):
    prob = sigmoid(sum(x * weights))
    if prob > 0.5: 
        return 1
    else: 
        return 0

绘制决策边界

# 绘图
def show(weights, x_train, y_train):
    x_train = np.array(x_train)
    n = np.shape(x_train)[0]
    # 坐标
    x1 = []
    y1 = []
    x2 = [] 
    y2 = []
    for i in range(n):
        if int(y_train[i]) == 1:
            x1.append(x_train[i, 1])
            y1.append(x_train[i, 2])
        else:
            x2.append(x_train[i, 1])
            y2.append(x_train[i, 2])
    fig = plt.figure()
    ax = fig.add_subplot(111)
    ax.scatter(x1, y1, s=30, c='red', marker='s')
    ax.scatter(x2, y2, s=30, c='green')
    x = np.arange(-3.0, 3.0, 0.1)
    weights = np.array(weights)
    y = (-weights[0] - weights[1] * x) / weights[2]
    ax.plot(x, y)
    plt.xlabel('X1')
    plt.ylabel('X2')
    plt.show()