python实现KMeans算法聚类分析

一，KMeans概述
KMeans算法将数据集中的数据点随机生成 k 组,把每组的均值作为 中心点。计算每个数据点与各组的中心点的相似性,根据数据点相似性的度量准 则,把每个数据点重新分组,计算每组新的均值作为中心点。不断重复上述过程,直 到中心点的均值收敛,停止迭代过程。

二，代码

from collections import Counter
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import random as rd


# Kmeans聚类
def Kmeans(k, data):
    # 随机选取k个样本点作为初始簇心
    cluster_center = data[rd.sample(list(range(len(data))), k)]
    # 对每个样本进行分类,初始都归为0类
    Cluster_tag = np.zeros((len(data),))
    # 迭代
    t = 10
    while t > 0:
        for i, sample in enumerate(data):
            # 样本与簇心的距离_distance
            _distance = distance(sample, cluster_center)
            # 得到该序列最小值的索引
            min_idx = np.argmin(_distance)
            if Cluster_tag[i] != min_idx:
                Cluster_tag[i] = min_idx
        # 更新簇心坐标
        for tag in range(k):
            centroid = np.mean(data[Cluster_tag == tag], axis=0)
            cluster_center[tag] = centroid
        t = t - 1
    print('质心坐标：\n', cluster_center)

    j = 0
    while j < len(range(k)):
        plt.scatter(cluster_center[j][0], cluster_center[j][1], c='red', marker='*', s=50)
        j = j + 1
    return Cluster_tag


# 导入数据
def Import_data():
    import_data = np.array(pd.read_csv(r'D:\Desktop\kmpractise\flame.txt', sep='\t', header=None, usecols=[0, 1]))
    return import_data


# 计算欧式距离
def distance(x, y):
    return np.sum((x - y) ** 2, axis=1) ** 0.5


# 聚类结果可视化展示
def show(data, labels):
    plt.scatter(data[:, 0], data[:, 1], c=labels, alpha=0.6, s=20)
    plt.grid(axis='both', linewidth=0.2)
    plt.show()


# 导出聚类结果
def Clustering_result(m, n):
    i = -1
    while i < n.shape[0] - 1:
        i = i + 1
        with open(r'D:\soft\PyCharm 2022.1.3\pythonproject\kms\answer.txt', 'a+', encoding='utf-8') as f:
            print(m[i][0], m[i][1], int(n[i]), sep='\t', file=f)


if __name__ == '__main__':
    data = Import_data()
    Cluster_tag = (Kmeans(2, data))
    dict1 = dict(Counter(Cluster_tag))
    print('0:', list(Cluster_tag).count(0), '1:', list(Cluster_tag).count(1))
    print('聚类标签：\n', Cluster_tag, '\n每类中的数据个数:\n', dict1)
    Clustering_result(data, Cluster_tag)
    ans = pd.read_csv(r'D:\soft\PyCharm 2022.1.3\pythonproject\kms\answer.txt', sep='\t', header=None)
    print('聚类结果展示(聚类结果已保存至answer.txt):\n', ans)

    show(data, Cluster_tag)

#要使用的数据集falme.txt（只展示部分数据）：

 

 

 

三，运行结果展示