k邻近算法（一）

简单的KNN邻近算法，如有问题，欢迎补充~

# -*- coding:utf-8 -*-                                                                                                                                                                                                                                          
'''
python3.6
简单的k邻近算法
'''
from numpy import *

# 抛出异常
def throw_error(err):
    raise Exception(err)

# 字典按value进行排序
def sortedDictValue(dict, reverse = False):
    arr = []
    if dict is None:
        return arr
    # 利用lambda表达式排序
    arr = sorted(dict.items(), key = lambda x:x[1], 
            reverse = reverse)      # 返回list
    return arr

# 创建数据集
def createDataSet():
    group = [ 
        [3, 104],
        [2, 100],
        [1, 81],
        [101, 10],
        [99, 5], 
        [98, 2], 
    ]   
    labels = ['A', 'A', 'A', 'B', 'B', 'B']
    return group, labels

# k邻近算法
def classifyKNN(vectorX, dataSet, labels, k):
    # 检查参数是否合法
    if not isinstance(labels, list):
        throw_error("labels必须为list数组")
    if not isinstance(k, int):
        throw_error("k参数必须为整数")

    # 训练集行数
    dataSize = dataSet.shape[0]
    k = dataSize if k > dataSize else k

    # 构造训练集与样本的差值矩阵
    diffMat = tile(vectorX, (dataSize, 1)) - dataSet

    # 求元素的平方
    sqDiffMat = power(diffMat, 2)

    # 行求和开方
    sumDiffMat = sqDiffMat.sum(axis=1)

    # 求出样本到每个训练集的距离
    distance = power(sumDiffMat, 0.5)

    # 获得从小到大的顺序并转换为一维数组
    sortDis = mat2OneArr(argsort(distance, axis=0))

    classCount = {}     # 训练集的计数
    for i in range(k):
        lb = labels[sortDis[i]]
        classCount[lb] = classCount.get(lb, 0) + 1
    # 根据距离大小排序
    sortClassCount = sortedDictValue(classCount, True)
    return sortClassCount[0][0]

def main():
    # 测试
    group, labels = createDataSet()
    try:
        ret = classifyKNN([0, 1], group, labels, 3);
        print(ret)
    except Exception as e:
        print(e)

if __name__ == '__main__':
    main()