iris.l1.cluster {directlabels}

Clustering of the iris data with the l1 clusterpath

Description

The l1 clustering algorithm from the clusterpath package was applied to the iris dataset and the breakpoints in the solution path are stored in this data frame.

Usage

data(iris.l1.cluster)

Format

A data frame with 9643 observations on the following 8 variables.

row

a numeric vector: row of the original iris data matrix

Species

a factor with levels setosa versicolor virginica: Species from corresponding row

alpha

a numeric vector: the value of the optimal solution.

lambda

a numeric vector: the regularization parameter (ie point in the path).

col

a factor with levels Sepal.Length Sepal.Width Petal.Length Petal.Width: column from the original iris data.

gamma

a factor with levels 0: parameter from clustering.

norm

a factor with levels 1 parameter from clustering.

solver

a factor with levels path algorithm used for clustering.

Source

clusterpath package

References

clusterpath article

Examples

data(iris.l1.cluster,package="directlabels")
iris.l1.cluster$y <- iris.l1.cluster$alpha
if(require(ggplot2)){
  p <- ggplot(iris.l1.cluster,aes(lambda,y,group=row,colour=Species))+
    geom_line(alpha=1/4)+
    facet_grid(col~.)
  p2 <- p+xlim(-0.0025,max(iris.l1.cluster$lambda))
  print(direct.label(p2,list(first.points,get.means)))
}