KNN算法尝试

KNN算法属于机器学习中的监督学习的分类问题的算法

KNN的原理就是当预测一个新的值x的时候，根据它距离最近的K个点是什么类别来判断x属于哪个类别

步骤：

1.计算已知类别数据集的点与当前点之间的距离

2.按照距离递增排序

3.选取与当前点距离最小的k个点

4.确定前K个点所在类别出现频率

5.返回出现最多的为当前类别

from numpy import *
import math

def get_dataset():
    group = array([[1, 1], [0, 0], [5, 5], [4, 10]])
    label = ['A', 'B', 'C', 'D']
    return group, label

def get_class(group, label, data, k):
    list = group
    dis = []
    index = []
    for item in group:
        a = array(item) - array(data)
        dis.append(math.hypot(a[0], a[1]))
    dis = array(dis)
    dis1 = dis.argsort()    # 按递减排序得到对应索引
    for i in range(k):
        index.append(dis1[i])
    count_a = 0
    count_b = 0
    count_c = 0
    count_d = 0
    for item in index:
        if label[item] == 'A':
            count_a += 1
        elif label[item] == 'B':
            count_b += 1
        elif label[item] == 'C':
            count_c += 1
        elif label[item] == 'D':
            count_d += 1
    if count_a == max(count_a, count_b, count_c, count_d):
        classname = 'A'
    elif count_b == max(count_a, count_b, count_c, count_d):
        classname = 'B'
    elif count_c == max(count_a, count_b, count_c, count_d):
        classname = 'C'
    elif count_d == max(count_a, count_b, count_c, count_d):
        classname = 'D'


if __name__ == '__main__':
    group, label = get_dataset()
    print(get_class(group, label, [2, 10], 1))