###################################################

Author : Kirthik D

K-Mean Clustering

###################################################

set.seed(256)

Load iris dataset into a new variable iris2

data(iris)

str(iris)

#Remove the initial label of Species from original dataset

iris2 <- iris[,-5]

iris2

Apply K-mean clustering to understand the Species from other attributes

?kmeans

kmeans.result <- kmeans(iris2, centers = 3, nstart = 20)

kmeans.result

See the cluster identification for each observation

kmeans.result$cluster

Compare with original label

table(iris$Species, kmeans.result$cluster)

Visualizing and interpreting results of k-means()

plot(iris2[c("Sepal.Length", "Sepal.Width")])

plot(iris2[c("Sepal.Length", "Sepal.Width")], col = kmeans.result$cluster)

plot(iris2[c("Sepal.Length", "Sepal.Width")], col = kmeans.result$cluster, main = "K-Means with 3 clusters")

plot cluster centers

kmeans.result$centers

kmeans.result$centers[,c("Sepal.Length", "Sepal.Width")]

points(kmeans.result$centers[,c("Sepal.Length", "Sepal.Width")], col = 1:3,pch = 8, cex=3)#cex = font size, pch = symbol

Visualising the clusters

library(cluster)

y_kmeans <- kmeans.result$cluster

?clusplot()

clusplot(iris2[,c("Sepal.Length", "Sepal.Width")],

     y_kmeans,

     lines = 0,

     shade = TRUE,

     color = TRUE,

     labels = 2,

     plotchar = FALSE,

     span = TRUE,

     main = paste('Clusters of iris'),

     xlab = 'Sepal.Length',

     ylab = 'Sepal.Width')

Determining number of clusters

See Total within sum of square error

kmeans.result$tot.withinss

Initialize total within cluster sum of squares error: wcss

set.seed(6)

wcss = vector()

For 1 to 15 cluster centers check the WCSS

for (i in 1:10) wcss[i] = sum(kmeans(iris2, centers = i,nstart = 20)$withinss)

Plot WSS vs. number of clusters

plot(1:10,

 wcss, 

 type = "b", 

 main = paste('The Elbow Method'),

 xlab = "Number of Clusters", 

 ylab = "WCSS")

Set k equal to the number of clusters corresponding to the elbow location

k <-3

Fitting K-Means to the dataset

set.seed(29)