R Programming - Clustering Algorithms

• 7th Sep, 2021
• 17:16 PM

This is the file19.txt we needed this file for calculating our problem HARTIGAN is a dataset directory that contains test data for clustering algorithms. The data files are all simple text files, and the format of the data files is explained on the web page

library(factoextra)
library(fpc)
library(dplyr)

#Q2.2 : K-means clustering (2.5 points divided evenly among the components)

clust<-read.csv("Cluster.csv")
str(clust)

cluste<-scale(clust[-1])
cluster<-data.frame(cluste)

rownames(cluster)<-clust$Name
res.dst<-get_dist(cluster, method="pearson")
fviz_dist(res.dst,lab_size=8)

res.km<-eclust(cluster, "kmeans",nstart=20)
fviz_gap_stat(res.km$gap_stat)

fviz_silhouette(res.km)

res.km$nbclust

fviz_nbclust(res.km)
fviz_cluster(res.km)

clusters<-res.km$cluster

Clust1<-cbind(cluster, clusters)
table(Clust1$clusters)

aggregate(clust[-1], by=list(cluster=res.km$cluster), mean)

#Q2.3 : Hierarchical clustering (3 points divided evenly among the components)

set.seed(1122)
clust_1<-sample(cluster,size=35, replace=TRUE)
rownames(clust_1)<-clust$Name

res.hclust<-eclust(clust_1, "hclust", hc_method="single")
fviz_dend(res.hclust, rect = TRUE)
fviz_silhouette(res.hclust)
fviz_cluster(res.hclust)
fviz_dend(res.hclust)

res.hclust_2<-eclust(clust_1[,2:9], "hclust", hc_method="complete")
fviz_dend(res.hclust_2, rect = TRUE)
fviz_silhouette(res.hclust_2)
fviz_cluster(res.hclust_2)
fviz_dend(res.hclust_2)

res.hclust_3<-eclust(clust_1[,2:9], "hclust", hc_method="average")
fviz_dend(res.hclust_3, rect = TRUE)
fviz_silhouette(res.hclust_3)
fviz_cluster(res.hclust_3)
fviz_dend(res.hclust_3)

#Complete produces least singleton sets
res.hclust_4<-eclust(clust_1[,2:9], "hclust",k=3, hc_method="complete")
fviz_dend(res.hclust_4, rect = TRUE)
fviz_silhouette(res.hclust_4)
fviz_cluster(res.hclust_4)
fviz_dend(res.hclust_4)