06-k-mediods-clust.Rmd

---
title: "K Mediods Clustering"
author: "Kundan K. Rao"
date: "01/12/2021"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE)
```

Loading required libraries:-
```{r}
library(cluster)
library(factoextra)
```

finding optimum number of cluster:-
```{r}
fviz_nbclust(data.scaled, pam, method = "silhouette")+
theme_classic()
```
Observation:-
optimum number of cluster is 3.

Finding clusters using k=3
```{r}
pam.res <- pam(data.scaled, 3)
print(pam.res)
```
Clustered data :-
```{r}
data.clust.medoids <- cbind(data, cluster = pam.res$cluster)
head(data.clust.medoids, n = 5)
```

Visualising clusters:-
```{r}
fviz_cluster(pam.res,
palette = c("#00AFBB", "#FC4E07","#00AB33"), # color palette
ellipse.type = "t", # Concentration ellipse
repel = TRUE, # Avoid label overplotting (slow)
ggtheme = theme_classic()
)
```

The clusters seems similar to k-means