【机器学习】聚类方法实例——使用Kmeans算法实现数据聚类分析

一、聚类目标

簇内相似度高&＃xff0c;且簇间相似度低&＃xff08;高内聚&＃xff0c;低耦合&＃xff09;&＃xff08;high-intra-class,low-inter-class&＃xff09;

1.K-Means(K均值)聚类

算法步骤&＃xff1a;

(1) 首先我们选择一些类/组&＃xff0c;并随机初始化它们各自的中心点。中心点是与每个数据点向量长度相同的位置。这需要我们提前预知类的数量(即中心点的数量)。
(2) 计算每个数据点到中心点的距离&＃xff0c;数据点距离哪个中心点最近就划分到哪一类中。
(3) 计算每一类中中心点作为新的中心点。
(4) 重复以上步骤&＃xff0c;直到每一类中心在每次迭代后变化不大为止。也可以多次随机初始化中心点&＃xff0c;然后选择运行结果最好的一个。

优点&＃xff1a;

速度快&＃xff0c;计算简便

缺点&＃xff1a;

必须提前知道数据有多少类/组。
K-Medians是K-Means的一种变体&＃xff0c;是用数据集的中位数而不是均值来计算数据的中心点。
K-Medians的优势是使用中位数来计算中心点不受异常值的影响&＃xff1b;缺点是计算中位数时需要对数据集中的数据进行排序&＃xff0c;速度相对于K-Means较慢。

二、使用步骤

1.生成数据

#导入需要的库
from numpy import unique
from numpy import where
from sklearn.datasets import make_classification
from sklearn.preprocessing import LabelEncoder, StandardScaler
from matplotlib import pyplot# 将生成的分类数据集用图形化显示
X,y &＃61; make_classification(n_samples&＃61;5000, n_features&＃61;2,n_informative&＃61;2, n_redundant&＃61;0,n_classes&＃61;3,n_clusters_per_class&＃61;1,random_state&＃61;1)
# 数据集标准化操作
X &＃61; StandardScaler().fit_transform(X)
clusters_temp &＃61; unique(y)
for cluster_temp in clusters_temp:row_ix &＃61; where(y &＃61;&＃61; cluster_temp)pyplot.scatter(X[row_ix,0],X[row_ix,1])
pyplot.show()#显示数据集分布情况
# y.shape


2.使用K-means算法进行聚类

from sklearn.cluster import KMeans
from matplotlib import pyplot
from numpy import unique
from numpy import where
from matplotlib import pyplotmodel &＃61; KMeans(n_clusters&＃61;3)#聚类类别为2
yhat &＃61; model.fit_predict(X)
clusters &＃61; unique(yhat)
for cluster in clusters:row_ix &＃61; where(yhat &＃61;&＃61; cluster)pyplot.scatter(X[row_ix,0],X[row_ix,1])
pyplot.show()


分类后是这个样子&＃xff1a;

3.评价聚类结果

from sklearn import metrics
from sklearn.metrics import davies_bouldin_score
print("Adjusted Rand-Index: %.3f" % metrics.adjusted_rand_score(y,yhat))
print("davies bouldin index: %.3f"% metrics.davies_bouldin_score(X,yhat))


评价结果如下&＃xff1a;

4.用silhouette选择K-means聚类的簇数&＃xff1a;

import numpy as np
from sklearn.cluster import KMeans
from sklearn.metrics import silhouette_samples, silhouette_scorerange_n_clusters &＃61; [2, 3, 4, 5, 6]for n_clusters in range_n_clusters:clusterer &＃61; KMeans(n_clusters&＃61;n_clusters, random_state&＃61;10)cluster_labels &＃61; clusterer.fit_predict(X)silhouette_avg &＃61; silhouette_score(X, cluster_labels)print("For n_clusters &＃61;",n_clusters,"The average silhouette_score is :",silhouette_avg,)


结果显示分成3簇的时候效果最好

5.使用不同聚类方法

from sklearn.cluster import Birchmodel &＃61; Birch(n_clusters&＃61;3)yhat &＃61; model.fit_predict(X)
clusters &＃61; unique(yhat)
for cluster in clusters:row_ix &＃61; where(yhat &＃61;&＃61; cluster)pyplot.scatter(X[row_ix,0],X[row_ix,1])
pyplot.show()print("The Birch adjusted_rand_score is:",metrics.adjusted_rand_score(y,yhat))
print("The Birch davies_bouldin_score is:",metrics.davies_bouldin_score(X,yhat))


结果如下&＃xff1a;

完整代码

#导入库
import numpy as np
from numpy import unique
from numpy import where
from matplotlib import pyplot
from sklearn.datasets import make_classification
from sklearn.preprocessing import LabelEncoder, StandardScaler
from sklearn.cluster import KMeans
from sklearn.metrics import silhouette_samples, silhouette_score# 生成数据集
X,y &＃61; make_classification(n_samples&＃61;5000, n_features&＃61;2,n_informative&＃61;2, n_redundant&＃61;0,n_classes&＃61;3,n_clusters_per_class&＃61;1,random_state&＃61;1)
X &＃61; StandardScaler().fit_transform(X)
################################################################################
range_n_clusters &＃61; [2, 3, 4, 5, 6]
for n_clusters in range_n_clusters: clusterer &＃61; KMeans(n_clusters&＃61;n_clusters, random_state&＃61;10)cluster_labels &＃61; clusterer.fit_predict(X)silhouette_avg &＃61; silhouette_score(X, cluster_labels)print("For n_clusters &＃61;",n_clusters,"The average silhouette_score is :",silhouette_avg,)
################################################################################
model &＃61; KMeans(n_clusters&＃61;3)#聚类类别为2
yhat &＃61; model.fit_predict(X)
clusters &＃61; unique(yhat)
for cluster in clusters:row_ix &＃61; where(yhat &＃61;&＃61; cluster)pyplot.scatter(X[row_ix,0],X[row_ix,1])
pyplot.show()
print("The KMeans Adjusted Rand-Index: %.3f" % metrics.adjusted_rand_score(y,yhat))
print("The KMeans davies bouldin index: %.3f"% metrics.davies_bouldin_score(X,yhat))
################################################################################
from sklearn.cluster import AgglomerativeClustering
model &＃61; AgglomerativeClustering(n_clusters&＃61;3)
yhat &＃61; model.fit_predict(X)
clusters &＃61; unique(yhat)
for cluster in clusters:row_ix &＃61; where(yhat &＃61;&＃61; cluster)pyplot.scatter(X[row_ix,0],X[row_ix,1])
pyplot.show()
print("The AgglomerativeClustering adjusted_rand_score is:",metrics.adjusted_rand_score(y,yhat))
print("The AgglomerativeClustering davies_bouldin_score is:",metrics.davies_bouldin_score(X,yhat))
################################################################################
from sklearn.cluster import SpectralClustering
model &＃61; SpectralClustering(n_clusters&＃61;3)
yhat &＃61; model.fit_predict(X)
clusters &＃61; unique(yhat)
for cluster in clusters:row_ix &＃61; where(yhat &＃61;&＃61; cluster)pyplot.scatter(X[row_ix,0],X[row_ix,1])
pyplot.show()
print("The SpectralClustering adjusted_rand_score is:",metrics.adjusted_rand_score(y,yhat))
print("The SpectralClustering davies_bouldin_score is:",metrics.davies_bouldin_score(X,yhat))
################################################################################
from sklearn.cluster import Birch
model &＃61; Birch(n_clusters&＃61;3)
yhat &＃61; model.fit_predict(X)
clusters &＃61; unique(yhat)
for cluster in clusters:row_ix &＃61; where(yhat &＃61;&＃61; cluster)pyplot.scatter(X[row_ix,0],X[row_ix,1])
pyplot.show()
print("The Birch adjusted_rand_score is:",metrics.adjusted_rand_score(y,yhat))
print("The Birch davies_bouldin_score is:",metrics.davies_bouldin_score(X,yhat))


结果如下&＃xff1a;

总结

以上是关于聚类方法实例的一种实现方法&＃xff0c;供大家参考。

参考

常见的六大聚类算法