08-heirarchical-clust.Rmd

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

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

# Hierarchical Clustering
Method in distance matrix:-
Euclidean
Linkage method in clustering algorithm:-
1. Complete
2. WARD.D2
```{r}
# hierarchical clustering
dist.mat <- dist(data.scaled, method = "euclidean")
hc.complete<-hclust(dist.mat,method = "complete")
hc.avg<-hclust(dist.mat,method = "average")
hc.ward<-hclust(dist.mat,method = "ward.D2")
```

```{r}
#str(as.dendrogram(hc.ward))
```

```{r}
# ISL book
# dendrograms plot
par(mfrow=c(1,2))
plot(hc.complete,main="Complete Linkage", xlab="",sub="",cex=.9)
plot(hc.ward,main="WARD D2", xlab="",sub="",cex=.9)
```


```{r}
# visualising dendograms
# clustering book
par(mfrow=c(1,2))
fviz_dend(hc.complete, cex = 0.5)
```
```{r}
fviz_dend(hc.eucl, cex = 0.5)
```
```{r}
fviz_dend(hc.avg, cex = .5)
```


Verify cluster tree:-
```{r}
# cophenetic distance
hc.complete.coph <- cophenetic(hc.complete)
hc.avg.coph <- cophenetic(hc.avg)
hc.ward.coph <- cophenetic(hc.ward)

# correlation b/w the cophenetic distance and original distance
cor(dist.mat, hc.complete.coph)
cor(dist.mat, hc.ward.coph)
cor(dist.mat, hc.avg.coph)
```

```{r}
# ISL book
# each wine's cluster
grp.complete <- cutree(hc.complete,3)
grp.avg <- cutree(hc.avg, 9)
grp.ward <- cutree(hc.ward,3)
table(grp.complete)
table(grp.avg)
table(grp.ward)
```


```{r}
grp.avg
grp.complete
grp.ward
```
Visualising cutree
```{r}
# Cut in 4 groups and color by groups
fviz_dend(hc.ward, k = 3, # Cut in four groups
cex = 0.5, # label size
k_colors = c("#2E66DF", "#00AFBB", "#E7B800"),
color_labels_by_k = TRUE, # color labels by groups
rect = TRUE, # Add rectangle around groups
rect_border = c("#2E66DF", "#00AFBB", "#E7B800"),
rect_fill = TRUE,
main = "Dendrogram - ward.D2",
xlab = "Objects", ylab = "Distance", sub = "",
ggtheme = theme_minimal()
)
```

```{r}
# Cut in 4 groups and color by groups
fviz_dend(hc.complete, k = 3, # Cut in four groups
cex = 0.5, # label size
k_colors = c("#2E00DF", "#55AFBB", "#E7B800"),
color_labels_by_k = TRUE, # color labels by groups
rect = TRUE # Add rectangle around groups
)
```

```{r}
# Cut in 4 groups and color by groups
fviz_dend(hc.avg, k = 9, # Cut in four groups
cex = 0.5, # label size
#k_colors = c("#2E9F00", "#00AFBB", "#E7B866"),
color_labels_by_k = TRUE, # color labels by groups
rect = TRUE # Add rectangle around groups
)
```
```{r}
# circular visualisation
fviz_dend(hc.ward, cex = 0.5, k = 3,
k_colors = "jco", type = "circular")
```
```{r}
library(igraph)
fviz_dend(hc.ward, k = 3, k_colors = "jco",
type = "phylogenic", repel = TRUE,
phylo_layout = "layout_as_tree"
)
```



As cluster:-
```{r}
fviz_cluster(list(data = data.scaled, cluster = grp.complete),
#palette = c("#2E9FDF", "#00AFBB", "#E7B800", "#FC4E07"),
ellipse.type = "convex", # Concentration ellipse
repel = TRUE, # Avoid label overplotting (slow)
show.clust.cent = FALSE, ggtheme = theme_minimal())
```

```{r}
fviz_cluster(list(data = data.scaled, cluster = grp.ward),
#palette = c("#2E9FDF", "#00AFBB", "#E7B800", "#FC4E07"),
ellipse.type = "convex", # Concentration ellipse
repel = TRUE, # Avoid label overplotting (slow)
show.clust.cent = FALSE, ggtheme = theme_minimal())
```
```{r}
fviz_cluster(list(data = data.scaled, cluster = grp.avg),
#palette = c("#2E9FDF", "#00AFBB", "#E7B800", "#FC4E07"),
ellipse.type = "convex", # Concentration ellipse
repel = TRUE, # Avoid label overplotting (slow)
show.clust.cent = FALSE, ggtheme = theme_minimal())
```

Using cluster package for hirarchical clustering
```{r}
library("cluster")
# Agglomerative Nesting (Hierarchical Clustering)
hc.agnes <- agnes(x = data, # data matrix
stand = TRUE, # Standardize the data
metric = "euclidean", # metric for distance matrix
method = "ward" # Linkage method
)
# DIvisive ANAlysis Clustering
hc.diana <- diana(x = data, # data matrix
stand = TRUE, # standardize the data
metric = "euclidean" # metric for distance matrix
)

fviz_dend(hc.agnes, cex = 0.6, k=3)
```
```{r}
fviz_dend(hc.diana, cex = 0.6, k=3)
```

Comparing dendograms:-
```{r}
library(dendextend)
# creating two dendograms
dend1 <- as.dendrogram(hc.avg)
dend2 <- as.dendrogram(hc.ward)

# creating a list to hold the dendograms
dend.list <- dendlist(dend1, dend2)
```

```{r}
tanglegram(dend1,dend2,
           highlight_distinct_edges = FALSE, # Turn-off dashed lines
common_subtrees_color_lines = FALSE, # Turn-off line colors
common_subtrees_color_branches = TRUE, # Color common branches
main = paste("entanglement =", round(entanglement(dend.list), 2))
)
```
```{r}
# Cophenetic correlation matrix
cor.dendlist(dend.list, method = "cophenetic")
```

```{r}
# Baker correlation matrix
cor.dendlist(dend.list, method = "baker")
```

```{r}
# Cophenetic correlation coefficient
cor_cophenetic(dend1, dend2)
```

```{r}
# Baker correlation coefficient
cor_bakers_gamma(dend1, dend2)
```