Selecting number of clusters

Unsupervised Learning in R

Hank Roark

Senior Data Scientist at Boeing

Interpreting results

# Create hierarchical cluster model: hclust.out
hclust.out <- hclust(dist(x))

# Inspect the result
summary(hclust.out)

            Length Class  Mode     
merge       98     -none- numeric  
height      49     -none- numeric  
order       50     -none- numeric  
labels       0     -none- NULL     
method       1     -none- character
call         2     -none- call     
dist.method  1     -none- character

Dendrogram

Tree shaped structure used to interpret hierarchical clustering models

five observations plotted and a dendrogram

Dendrogram

Tree shaped structure used to interpret hierarchical clustering models

two cluster points are joined in the dendrogram

Dendrogram

Tree shaped structure used to interpret hierarchical clustering models

two other cluster points are joined in the dendrogram

Dendrogram

Tree shaped structure used to interpret hierarchical clustering models

two clusters and a point are joined in the dendrogram

Dendrogram

Tree shaped structure used to interpret hierarchical clustering models

all clusters and points are joined together in the dendrogram

Dendrogram plotting in R

# Draws a dendrogram
plot(hclust.out)

abline(h = 6, col = "red")

horizontal bar showing how many clusters we want on the dendrogram

Tree "cutting" in R

# Cut by height h
cutree(hclust.out, h = 6)

1 1 1 1 1 1 1 1 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 3 3 3
3 3 3 3 3 3 3 4 4 4 4 4 4 4 4 4 4 2 4 2 4 4

# Cut by number of clusters k
cutree(hclust.out, k = 2)

1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2
2 2 2 2 2 2 2 1 1 1 1 1 1 1 1 1 2 1 1 1 1 1

Let's practice!

Unsupervised Learning in R