ML: 聚类算法R包 - 模型聚类

模型聚类

---

• mclust::Mclust
• RWeka::Cobweb

mclust::Mclust

---

EM算法也称为期望最大化算法，在是使用该算法聚类时，将数据集看作一个有隐形变量的概率模型，并实现模型最优化，即获取与数据本身性质最契合的聚类方式为目的，通过‘反复估计’模型参数找出最优解，同时给出相应的最有类别级数k

所需程序安装包

install.packages("mclust")

函数示例代码

> library(mclust)
>  EM<-Mclust(iris[,-5])
>  summary(EM,parameters=T)
----------------------------------------------------
Gaussian finite mixture model fitted by EM algorithm
----------------------------------------------------

Mclust VEV (ellipsoidal, equal shape) model with 2 components:

 log.likelihood   n df       BIC       ICL
       -215.726 150 26 -561.7285 -561.7289

Clustering table:
  1   2
 50 100 

Mixing probabilities:
       1        2
0.333332 0.666668 

Means:
                  [,1]     [,2]
Sepal.Length 5.0060021 6.261996
Sepal.Width  3.4280046 2.871999
Petal.Length 1.4620006 4.905993
Petal.Width  0.2459998 1.675997

可以看到最优类别级数为2，各类分别含有50，100，

mclust::plot.Mclust(EM,what = "classification")

mclust::plot.Mclust(EM,what = "density")

RWeka::Cobweb

---

COBWEB是一种流行的简增量概念聚类算法。它以一个分类树的形式创建层次聚类，每个节点对应一个概念，包含该概念的一个概率描述，概述被分在该节点下的对象。使得该函数，需要安装RWeka包，在安装的过程中，可能出现如下的异常

** R
** inst
** preparing package for lazy loading
Error : .onLoad failed in loadNamespace() for 'rJava', details:
  call: fun(libname, pkgname)
  error: JAVA_HOME cannot be determined from the Registry
ERROR: lazy loading failed for package 'RWeka'
* removing 'C:/Users/zhushy/Documents/R/win-library/3.2/RWeka'

安装jre, 参考资料： http://blog.csdn.net/afei__/article/details/51464783

RWeka包未安装成功，示例代码未验证，待确认

library(RWeka)

dcom=iris[,-5]
c1<-Cobweb(dcom)
c1
c1$class_ids
table(predict(c1),dcom$clas)

参考资料：

• https://cran.r-project.org/web/packages/mclust/index.html
• https://www.douban.com/note/495288321/