K-Means算法学习

一、K-Means代码实现

import numpy as np
import matplotlib.pyplot as plt

# 加载数据
def loadDataSet(fileName):
    data = np.loadtxt(fileName,delimiter=',')
    return data

# 欧式距离计算
def distEclud(x,y):
    return np.sqrt(np.sum((x-y)**2))

# 为给定数据集构建一个包含K个随机质心的集合
def randCent(dataSet,k):
    # m为行数，n为列数
    m,n = dataSet.shape
    centroids = np.zeros((k,n))
    for i in range(k):
        index = int(np.random.uniform(0,m))
        centroids[i,:] = dataSet[index,:]
    return centroids

# k均值聚类
def KMeans(dataSet,k):
    # 行的数目
    m = np.shape(dataSet)[0]
    # 第一列存样本属于哪个簇   第二列存样本到簇的中心点的误差
    clusterAssment = np.mat(np.zeros((m,2)))
    clusterChange = True

    # 第一步：初始化centroids
    centroids = randCent(dataSet,k)
    while clusterChange:
        clusterChange = False
        # 遍历样本
        for i in range(m):
            minDist = 100000.0
            minIndex = -1
            # 遍历质心
            # 第二步 找出最近的质心
            for j in range(k):
                # 计算欧式距离
                distance = distEclud(centroids[j,:],dataSet[i,:])
                if distance < minDist:
                    minDist = distance
                    minIndex = j
            # 第三步 更新每行样本所属的簇
            if clusterAssment[i,0] != minIndex:
                clusterChange = True
                clusterAssment[i,:] = minIndex,minDist**2
        # 第四步 更新质心
        for j in range(k):
            # 获取聚类所有的点
            pointInCluster = dataSet[np.nonzero(clusterAssment[:,0].A == j)[0]]
            # 对矩阵的行求均值
            centroids[j,:] = np.mean(pointInCluster,axis=0)
    print("Congratulations,cluster complete!")
    return centroids,clusterAssment

def showCluster(dataSet,k,centroids,clusterAssment):
    m,n = dataSet.shape
    if n != 2:
        print("数据不是二维的")
        return 1
    mark = ['or', 'ob', 'og', 'ok', '^r', '+r', 'sr', 'dr', '<r', 'pr']
    if k > len(mark):
        print("k值太大了")
        return 1
    # 绘制样本
    for i in range(m):
        markIndex = int(clusterAssment[i,0])
        plt.plot(dataSet[i,0],dataSet[i,1],mark[markIndex])
    mark = ['Dr', 'Db', 'Dg', 'Dk', '^b', '+b', 'sb', 'db', '<b', 'pb']
    # 绘制质心
    for i in range(k):
        plt.plot(centroids[i,0],centroids[i,1],mark[i])
    plt.show()
dataSet = loadDataSet("test.txt")
k = 4
centroids,clusterAssment = KMeans(dataSet,k)
showCluster(dataSet,k,centroids,clusterAssment)

二、用sklearn.cluster带的K-Means算法实现

import numpy as np
import matplotlib.pyplot as plt
from sklearn.cluster import KMeans
# 加载数据
def loadDataSet(fileName):
    data = np.loadtxt(fileName,delimiter=',')
    return data
x = loadDataSet("test.txt")
y_pred = KMeans(n_clusters=3, random_state=9).fit_predict(x)
plt.scatter(x[:, 0], x[:, 1], c=y_pred)
plt.show()

三、K-Means效果的评价指标

评价指标：Silhouette Coefficient和Calinski-Harabasz Index
（1）Silhouette Coefficient分数值s的数学计算公式为：

a(i)是i到同簇样本的平均距离，为样本i的簇内不相似度，b(i)为i到其他簇样本的平均距离，为样本i的簇间不相似度。
s(i)接近1，样本在该簇合理；s(i)接近-1，样本i应该分到其他簇。

（2）Calinski-Harabasz分数值s的数学计算公式是：
s(k)=tr(Bk)(m-k)/tr(Wk)(k−1)*

其中m为训练集样本数，k为类别数。Bk为类别之间的协方差矩阵，Wk为类别内部数据的协方差矩阵。tr为矩阵的迹。类间差距越大，类内差距越小，s越大，聚类效果越好。

代码实现：

import numpy as np
import matplotlib.pyplot as plt
from sklearn.cluster import KMeans
from sklearn import metrics
# 加载数据
def loadDataSet(fileName):
    data = np.loadtxt(fileName,delimiter=',')
    return data

x = loadDataSet("test.txt")
y_pred = KMeans(n_clusters=3, random_state=9).fit_predict(x)
caha=metrics.calinski_harabasz_score(x, y_pred)
print(caha)
plt.scatter(x[:, 0], x[:, 1], c=y_pred)
plt.show()

借鉴：
SKlearn里面的K-means使用详解
关于K-means聚类算法重点理解