DBSCAN

DBSCAN

DBSCAN is a density based clustering algorithm. To use it we install the library dbscan and load it:

library(dbscan)
data <- iris[,-5]
clustering = dbscan(data, eps = 0.5, minPts = 5)
clustering

## DBSCAN clustering for 150 objects.
## Parameters: eps = 0.5, minPts = 5
## The clustering contains 2 cluster(s) and 17 noise points.
## 
##  0  1  2 
## 17 49 84 
## 
## Available fields: cluster, eps, minPts

The algorithm returns the clsuters and whether there are any noise points. The noise points are denoted with 0. To get the groupings:

clustering$cluster

##   [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 1 1 1 1 1 1 1 1 1 1 1
##  [36] 1 1 1 1 1 1 0 1 1 1 1 1 1 1 1 2 2 2 2 2 2 2 0 2 2 0 2 2 2 2 2 2 2 0 2
##  [71] 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 0 2 2 2 2 2 0 2 2 2 2 0 2 2 2 2 2 2
## [106] 0 0 2 0 0 2 2 2 2 2 2 2 0 0 2 2 2 0 2 2 2 2 2 2 2 2 0 2 2 0 0 2 2 2 2
## [141] 2 2 2 2 2 2 2 2 2 2

The classes and clusters comparison:

table(clustering$cluster, iris$Species)

##    
##     setosa versicolor virginica
##   0      1          6        10
##   1     49          0         0
##   2      0         44        40

Another package that contains dbscan is the fpc package.

An example specialized for DBSCAN

Lets create and plot the dataset

x1 <- rnorm(1000,0,3)
y1 <- rnorm(1000,-5,2)
x2 <- runif(1000, -20, 20)
y2 <- -0.05*x2^2 + 10 + runif(1000, -1, 1)
y <- c(y1,y2)
x <- c(x1,x2)
plot(x,y)

demo <- data.frame(x,y)
kmeansClust <- kmeans(demo, 2)
plot(demo, col = kmeansClust$cluster + 1, main="kmeans")

dbscanClust <- dbscan(demo, eps = 1, minPts = 12)
plot(demo, col = dbscanClust$cluster + 1, pch = ifelse(dbscanClust$cluster, 1, 4), main="dbscan")