机器学习实战KNN部分代码改写

首先创建测试样例

def createData():
    group = array([[1.1,1.1],[1.0,1.0],[0,0],[0,0.1]])
    labels = ['A','A','B','B']
    return group,labels

原书核心代码

def classify(inX,dataset,labels,k):
    datasetsize = dataset.shape[0]
    diffmat = tile(inX,(datasetsize,1)) - dataset
    #print(diffmat)
    sqdiffmat = diffmat**2
    #print(sqdiffmat)
    sqdistance = sqdiffmat.sum(axis=1)
    distances = sqdistance ** 0.5
    sorteddist = distances.argsort()
    classcount = {}
    for i in range(k):
        votelabels = labels[sorteddist[i]]
        #print(votelabels)
        classcount[votelabels] = classcount.get(votelabels,0) + 1
        #print(classcount)
    sorteddist = sorted(classcount.items(),key=operator.itemgetter(1),reverse=True)
    return sorteddist[0][0]

原函数中使用的大量的numpy方法，本文将用一些较为基础的代码对其中的方法进行简单的改写

一.tile方法改写

tile方法用于重复扩充行列

                                        

 将其改写为如下（需要控制输入格式，a为矩阵，*args为两个元素的元组）：

def tile1(a,*args):
    new_a = a
    x= args[0][0]
    y = args[0][1]
    for i in range(y-1):
        for j in range(len(a)):
            new_a.append(a[j])
    new_new = []
    for c in range(x):
        new_new.append(new_a)
    return new_new

二.改写array.sum(axis=1)

array.sum用于将矩阵的每一行值相加

将其改写为

def add_mat(a):
    mat = []
    for i in range(len(a)):
        sum = 0
        for j in range(len(a[0])):
            sum = sum + a[i][j]
        mat.append(sum)
    mat = array(mat)
    return mat        

 三.改写numpy.argsort

array.argsort是先排序（默认从小到大），而后返回排序后个元素在原数组的索引

将其改写为

def sorted_1(distances):
    sorted_list = []
    for i in range(len(distances)):
        sort_list = []
        sort_list.append(distances[i])
        sort_list.append(i)
        sorted_list.append(sort_list)
    #print(sorted_list)
    for i in range(len(sorted_list)):
        for i in range(0,len(sorted_list)-i-1):
            if(sorted_list[i][0] > sorted_list[i+1][0]):
                sorted_list[i],sorted_list[i+1] = sorted_list[i+1],sorted_list[i]
    new_list = []
    for i in range(len(sorted_list)):
        new_list.append(sorted_list[i][1])
    new_list = array(new_list)
    return new_list

四.改写sorted函数

sorted() 作为内置函数之一，其功能是对序列（列表、元组、字典、集合、还包括字符串）进行排序

 将其改写为

def sorted_dict(dict_1):
    dict_to_list = []
    #先将其改写为列表
    for key in dict_1:
        dict_list = []
        dict_list.append(key)
        dict_list.append(dict_1[key])
        dict_to_list.append(dict_list)
    #再对列表进行排序
    for i in range(len(dict_to_list)):
        for i in range(0,len(dict_to_list)-i-1):
            if(dict_to_list[i][0] < dict_to_list[i+1][0]):
                dict_to_list[i],dict_to_list[i+1] = dict_to_list[i+1],dict_to_list[i]
    return dict_to_list

主函数

def classify_0(x,dataset,labels,k):
    datasetsize = dataset.shape[0]
    diffmat = tile1(x,(datasetsize,1)) - dataset
    #print(diffmat)
    sqdiffmat = diffmat**2
    #print(sqdiffmat)
    
    sqdistance = add_mat(sqdiffmat)
    distances = sqdistance ** 0.5
    print(distances)
    sorteddist = sorted_1(distances)
    #print(sorteddist)
    #定义字典存储A类和B类的数量
    classcount = {}
    for i in range(k):
        votelabels = labels[sorteddist[i]]
        #print(votelabels)
        classcount[votelabels] = classcount.get(votelabels,0) + 1
        #print(classcount)
    sorteddist = sorted_dict(classcount)
    return sorteddist[0][0]

 输出结果为：