K-Means聚类算法及Python实现

Repeat {
for i = 1 to m
c(i) := index (form 1 to K) of cluster centroid closest to x(i)
for k = 1 to K
μk := average (mean) of points assigned to cluster k
}

from sklearn.cluster import KMeans  # 导入 sklearn.cluster.KMeans 类
import numpy as np
 
X = np.array([[1,2], [1,4], [1,0], [10,2], [10,4], [10,0]])
kmCluster = KMeans(n_clusters=2).fit(X)  # 建立模型并进行聚类，设定 K=2
print("聚类中心坐标:",kmCluster.cluster_centers_)  # 返回每个聚类中心的坐标
print("分类结果:",kmCluster.labels_)  # 返回样本集的分类结果
print("显示预测判断:",kmCluster.predict([[0, 0], [12, 3]]))  # 根据模型聚类结果进行预测判断

聚类中心坐标: [[10.  2.]
 [ 1.  2.]]
分类结果: [1 1 1 0 0 0]
显示预测判断: [1 0]
 
Process finished with exit code 0

from sklearn.cluster import MiniBatchKMeans # 导入 .MiniBatchKMeans 类
import numpy as np
X = np.array([[1,2], [1,4], [1,0], [4,2], [4,0], [4,4],
              [4,5], [0,1], [2,2],[3,2], [5,5], [1,-1]])
# fit on the whole data
mbkmCluster = MiniBatchKMeans(n_clusters=3,batch_size=6,max_iter=10).fit(X)
print("聚类中心的坐标:",mbkmCluster.cluster_centers_) # 返回每个聚类中心的坐标
print("样本集的分类结果:",mbkmCluster.labels_)  # 返回样本集的分类结果
print("显示判断结果：样本属于哪个类别:",mbkmCluster.predict([[0,0], [4,5]]))  # 根据模型聚类结果进行预测判断

聚类中心的坐标: [[ 2.55932203  1.76271186]
 [ 0.75862069 -0.20689655]
 [ 4.20588235  4.5       ]]
样本集的分类结果: [0 0 1 0 0 2 2 1 0 0 2 1]
显示判断结果：样本属于哪个类别: [1 2]
 
Process finished with exit code 0

from sklearn.cluster import kmeans_plusplus
from sklearn.datasets import make_blobs
import matplotlib.pyplot as plt
 
# Generate sample data
n_samples = 4000
n_components = 4
 
X, y_true = make_blobs(
    n_samples=n_samples, centers=n_components, cluster_std=0.60, random_state=0
)
X = X[:, ::-1]
 
# Calculate seeds from kmeans++
centers_init, indices = kmeans_plusplus(X, n_clusters=4, random_state=0)
 
# Plot init seeds along side sample data
plt.figure(1)
colors = ["#4EACC5", "#FF9C34", "#4E9A06", "m"]
 
for k, col in enumerate(colors):
    cluster_data = y_true == k
    plt.scatter(X[cluster_data, 0], X[cluster_data, 1], c=col, marker=".", s=10)
 
plt.scatter(centers_init[:, 0], centers_init[:, 1], c="b", s=50)
plt.title("K-Means++ Initialization")
plt.xticks([])
plt.yticks([])
plt.show()

#  -*- coding: utf-8 -*-
import numpy as np
import pandas as pd
from sklearn.cluster import KMeans, MiniBatchKMeans
 
def main():
    # 读取数据文件
    file = pd.read_excel('K-Means.xlsx', header=0)  # 首行为标题行
    file = file.dropna()  # 删除含有缺失值的数据
    # print(file.dtypes)  # 查看 df 各列的数据类型
    # print(file.shape)  # 查看 df 的行数和列数
    print(file.head())
 
    # 数据准备
    z_scaler = lambda x:(x-np.mean(x))/np.std(x)  # 定义数据标准化函数
    dfScaler = file[['D1','D2','D3','D4','D5','D6','D7','D8','D9','D10']].apply(z_scaler)  # 数据归一化
    dfData = pd.concat([file[['地区']], dfScaler], axis=1)  # 列级别合并
    df = dfData.loc[:,['D1','D2','D3','D4','D5','D6','D7','D8','D9','D10']]  # 基于全部 10个特征聚类分析
    # df = dfData.loc[:,['D1','D2','D7','D8','D9','D10']]  # 降维后选取 6个特征聚类分析
    X = np.array(df)  # 准备 sklearn.cluster.KMeans 模型数据
    print("Shape of cluster data:", X.shape)
 
    # KMeans 聚类分析(sklearn.cluster.KMeans)
    nCluster = 4
    kmCluster = KMeans(n_clusters=nCluster).fit(X)  # 建立模型并进行聚类，设定 K=4
    print("Cluster centers:\n", kmCluster.cluster_centers_)  # 返回每个聚类中心的坐标
    print("Cluster results:\n", kmCluster.labels_)  # 返回样本集的分类结果
 
    # 整理聚类结果(太棒啦！)
    listName = dfData['地区'].tolist()  # 将 dfData 的首列 '地区' 转换为 list
    dictCluster = dict(zip(listName,kmCluster.labels_))  # 将 listName 与聚类结果关联，组成字典
    listCluster = [[] for k in range(nCluster)]
    for v in range(0, len(dictCluster)):
        k = list(dictCluster.values())[v]  # 第v个城市的分类是 k
        listCluster[k].append(list(dictCluster.keys())[v])  # 将第v个城市添加到 第k类
    print("\n聚类分析结果(分为{}类):".format(nCluster))  # 返回样本集的分类结果
    for k in range(nCluster):
        print("第 {} 类：{}".format(k, listCluster[k]))  # 显示第 k 类的结果
 
    return
 
if __name__ == '__main__':
    main()

#*========================1. 建立工程，导入sklearn相关包======================================**
 
import numpy as np
from sklearn.cluster import KMeans
 
#*======================2. 加载数据，创建K-means算法实例，并进行训练，获得标签====================**
 
def loadData(filePath):
    fr = open(filePath, 'r+')            #r+：读写打开一个文本文件
    lines = fr.readlines()           #.readlines() 一次读取整个文件（类似于 .read() ) .readline() 每次只读.readlines() 慢得多。
    retData = []                     #retData：用来存储城市的各项消费信息
    retCityName = []                 #retCityName：用来存储城市名称
    for line in lines:
        items = line.strip().split(",")
        retCityName.append(items[0])
        retData.append([float(items[i]) for i in range(1, len(items))])
    return retData, retCityName      #返回值：返回城市名称，以及该城市的各项消费信息
 
def main():
    data, cityName = loadData('city.txt')    #1.利用loadData方法读取数据
    km = KMeans(n_clusters=4)                #2.创建实例
    label = km.fit_predict(data)             #3.调用Kmeans（）fit_predict()方法进行计算
    expenses = np.sum(km.cluster_centers_, axis=1)
    # print(expenses)
    CityCluster = [[], [], [], []]          #将城市按label分成设定的簇
    for i in range(len(cityName)):
        CityCluster[label[i]].append(cityName[i])   #将每个簇的城市输出
    for i in range(len(CityCluster)):              #将每个簇的平均花费输出
        print("Expenses:%.2f" % expenses[i])
        print(CityCluster[i])
 
if __name__ == '__main__':
    main()
 
#*=============3. 输出标签，查看结果========================================**
 
#将城市按照消费水平n_clusters类，消费水平相近的城市聚集在一类中
#expense：聚类中心点的数值加和，也就是平均消费水平