《机器学习实战》python3代码（5）：逻辑回归

逻辑回归的优缺点

优点：

1.容易实现
2.计算量小。对比于knn,实在是太小了

缺点：

1.容易欠拟合
2.处理特征空间很大，多类特征数据的时候不方便，处理非线性特征需要转换

逻辑回归思想

梯度上升

数学原理如下图



基本思想是通过一定次数的梯度上升（或者下降）输出经过数次调整的θ矩阵

逻辑回归梯度上升算法的构建

步骤一

数据预处理，将数据和标签分离

步骤二

确定循环次数，通过上述推导的梯度上升数学公式不断更新θ矩阵，随后输出

代码

分离数据和标签的代码

def loadDataSet():
    dataMat = []
    labelMat = []
    fr = open("D:/machine learning actual combat/testSet.txt")
    for line in fr.readlines():
        lineArr = line.strip().split()
        dataMat.append([1.0, float(lineArr[0]), float(lineArr[1])])
        labelMat.append(int(lineArr[2]))
    return dataMat, labelMat

sigmoid函数

def sigmoid(inX):
    return 1.0/(1+exp(-inX))

循环迭代生成权重矩阵（即θ）的函数

def gradAscent(dataMatIn, classLabels):
    dataMatrix = mat(dataMatIn)
    labelMat = mat(classLabels).transpose()
    m, n = shape(dataMatrix)

    alpha = 0.001
    maxCycles = 500
    weights = ones((n, 1))
    #print(weights)
    for k in range(maxCycles):
        h = sigmoid(dataMatrix * weights)# 100*1的列向量
        err = (labelMat - h)
        weights = weights + alpha * dataMatrix.transpose() * err
    return weights

此处将权重矩阵初始值都设为1
h即为推导过程中的hθ函数
而xθ对θ求导之后变成它的转置矩阵
故 按照梯度上升的推导公式，weights = weights + alpha * dataMatrix.transpose() * err（alpha为步长，其实此处的alpha为alpha1/m）

画出决策边界的函数

def plotBestFit(wei):
    import matplotlib.pyplot as plt
    #weights = wei.getA()#将矩阵转化为数组
    weights = wei
    dataMat, labelMat = loadDataSet()
    dataArr = array(dataMat)
    n = shape(dataArr)[0]
    xcord1 = []; ycord1 = []
    xcord2 = []; ycord2 = []
    for i in range(n):
        if int(labelMat[i]) == 1:
            xcord1.append(dataArr[i,1]); ycord1.append(dataArr[i,2])
        else: xcord2.append(dataArr[i,1]); ycord2.append(dataArr[i,2])
    fig = plt.figure()
    ax = fig.add_subplot(111)
    ax.scatter(xcord1, ycord1, s = 30, c = 'red', marker='s')
    ax.scatter(xcord2, ycord2, s = 30, c = 'green')
    x = arange(-3.0, 3.0, 0.1)
    y = (-weights[0]- weights[1]*x)/weights[2]#化线
    ax.plot(x, y)
    plt.xlabel('X1');
    plt.ylabel('X2');
    plt.show()

这里要特别注意！完全按照书上的代码总会出现错误。
如果输入的权重（weight为一个数组，需要对其作getA函数处理），比如书上未改进的梯度上升代码
而如果是一个列表就不需要

改进：随机梯度上升

def stocGradAscent0(dataMatrix, classLabels):
    m, n = shape(dataMatrix)
    alpha = 0.01
    weights = ones(n)
    for i in range(m):
        h = sigmoid(sum(dataMatrix[i]*weights))
        err = classLabels[i] - h
        weights = weights + alpha * err * dataMatrix[i]
    return weights

weights2 = stocGradAscent0(array(dataArr), labelMat)

此处的回归系数不再是数组，而是一个列表。
梯度上升每次更新回归系数时都会用上所有数据集，如果数据集太大就会不方便所以就有了随机梯度上升，一次更新只用一个
对比于梯度上升，随机梯度上升计算量小，且更容易达到全局最值，但是不容易收敛

于是就有了改进的随机梯度上升

改进的随机梯度上升

def stocGradAscent1(dataMatrix, classLabels, numIter = 150):
    m, n = shape(dataMatrix)
    weights = ones(n)
    for j in range(numIter):
        dataIndex = list(range(m))
        for i in range(m):
            alpha = 4/(1.0 + i + j) + 0.01
            randIndex = int(random.uniform(0, len(dataIndex)))
            h = sigmoid(sum(dataMatrix[randIndex] * weights))
            err = classLabels[randIndex] - h
            weights = weights + alpha * err * dataMatrix[randIndex]
            del(dataIndex[randIndex])
    return weights

步长alpha每次迭代都在减小，却不会小于0
同时，随机选择样本，而不是根据规律来，可以减少周期性的波动（由于某些值不是线性可分造成）

实战代码：预测病马

def classifyVector(inX ,weights):
    prob = sigmoid(sum(inX *weights))
    if prob > 0.5: return 1.0
    else: return 0.0

def colicTest():
    frTrain = open("D:/machine learning actual combat/horseColicTraining.txt")
    frTest = open("D:/machine learning actual combat/horseColicTest.txt")
    trainingSet = [];trainingLabels = []
    for line in frTrain.readlines():
        currentLine = line.strip().split('\t')
        lineArr = []
        for i in range(21):
            lineArr.append(float(currentLine[i]))
        trainingSet.append(lineArr)
        trainingLabels.append(float(currentLine[21]))
    trainWeights = stocGradAscent1(array(trainingSet), trainingLabels, 500)
    errCount = 0.0
    numTestVec = 0.0
    for line in frTest.readlines():
        numTestVec += 1
        currentLine = line.strip().split('\t')
        lineArr = []
        for i in range(21):
            lineArr.append(float(currentLine[i]))
        if int(classifyVector(array(lineArr), trainWeights)) != int(currentLine[21]):
            errCount += 1
    errRate = float(errCount)/numTestVec
    print('the errRate = %s' % errRate)
    return errRate

def multiTest():
    numTest = 10
    errSum = 0.0
    for key in range(numTest):
        errSum += colicTest()
    print('after %d iterations the average error rate is %f'%(numTest, errSum/float(numTest)))

最后

本书提供的数据集，可到github下载
或参考此博客：https://blog.csdn.net/sinat_29957455/article/details/79123394

大二小白一枚，欢迎指正也欢迎讨论