python实现K-Means算法

python实现K-Means算法

k-means算法的实现原理就不再赘述，给大家说一下程序的大致思路。

程序定义了两个方法，一个是计算欧氏距离（也就是两点之间线段最短，用勾股定理求斜边的长度）一个就是冗余很大的均值junzhi方法，这个方法实现了算法中的求均值、求每次更新的聚类中心步骤，方法最后使用if条件判断程序的出口
if origin_center_data_new != origin_center_data: junzhi(origin_data, origin_center_data_new)
代码块的最后是程序的入口，提供了8对数据origin_data并且使用给定的初始聚类中心origin_center_data，这里还没有考虑随机生成聚类中心。
origin_data = [[2, 10], [2, 5], [8, 4], [5, 8], [7, 5], [6, 4], [1, 2], [4, 9]] origin_center_data = [[2, 10], [5, 8], [1, 2]] junzhi(data=origin_data, center=origin_center_data)

程序不是非常困难，没有标注释，我也许会有一天感慨道：这是我自己写的吗？ ，需要耐心看完。嘿嘿

from math import sqrt
from numpy import *
# 计算欧氏距离

def o_dis(vector1, vector2):
    return sqrt(pow((vector1[0] - vector2[0]), 2) + pow((vector1[1] - vector2[1]), 2))


def junzhi(data, center):
    distence_virtul = 100
    origin_data = data
    origin_center_data = center
    group0 = []
    group1 = []
    group2 = []
    distence = []
    center_group = []
    for i in range(0, int((sum([len(arr) for arr in origin_data]) / 2))):
        for j in range(0, int((sum([len(arr) for arr in origin_center_data]) / 2))):
            center_data_iterator = origin_center_data[j]
            origin_data_iterator = origin_data[i]
            print("原始数据为" + str(origin_data_iterator), end='')
            print("中心点为" + str(center_data_iterator), end='')
            center_group.append(center_data_iterator)
            # 计算最小距离
            distence_real = o_dis(origin_data_iterator, center_data_iterator)
            print("距离为：" + str(distence_real))
            distence.append(distence_real)
        print(distence)
        print(center_group)
        for m in range(0, len(distence)):
            distence_real = distence[m]
            if (distence_real <= distence_virtul):
                distence_virtul = distence_real
            i = -1
            for x in distence:
                i = i + 1
                if str(distence_virtul) == (str(x)):
                    index = i
                    break
        print("最小距离为" + str(distence_virtul))
        print("最小距离对应的中心点为" + str(center_group[i]))
        if (center_group[i] == origin_center_data[0]):
            group0.append(origin_data_iterator)
        elif (center_group[i] == origin_center_data[1]):
            group1.append(origin_data_iterator)
        else:
            group2.append(origin_data_iterator)
        print(group0)
        print(group1)
        print(group2)
        print("-------------------------------------------------------------------------")
        distence_virtul = 100
        distence.clear()
        center_group.clear()
    # 计算均值，求出下一次的中心点
    total_x = 0
    total_y = 0
    for i in range(0, len(group0)):
        total_x = total_x + group0[i][0]
        total_y = total_y + group0[i][1]
    new_center_1 = [int(total_x / len(group0)), int(total_y / len(group0))]
    print(new_center_1)
    total_x = 0
    total_y = 0
    for i in range(0, len(group1)):
        total_x = total_x + group1[i][0]
        total_y = total_y + group1[i][1]
    new_center_2 = [int(total_x / len(group1)), int(total_y / len(group1))]
    print(new_center_2)
    total_x = 0
    total_y = 0
    for i in range(0, len(group2)):
        total_x = total_x + group2[i][0]
        total_y = total_y + group2[i][1]
    new_center_3 = [int(total_x / len(group2)), int(total_y / len(group2))]
    print(new_center_3)
    origin_center_data_new = [new_center_1, new_center_2, new_center_3]
    print("-------------------------------------------------------------------------")
    print(
        "下一次的中心点" + str(origin_center_data_new) + "==================================================================")
    if origin_center_data_new != origin_center_data:
        junzhi(origin_data, origin_center_data_new)
    return origin_center_data_new


origin_data = [[2, 10], [2, 5], [8, 4], [5, 8], [7, 5], [6, 4], [1, 2], [4, 9]]
origin_center_data = [[2, 10], [5, 8], [1, 2]]
junzhi(data=origin_data, center=origin_center_data)