| M5data {tclust} | R Documentation |
M5data data
Description
A bivariate data set obtained from three normal bivariate distributions with different scales and proportions 1:2:2. One of the components is very overlapped with another one. A 10% background noise is added uniformly distributed in a rectangle containing the three normal components and not very overlapped with the three mixture components. A precise description of the M5 data set can be found in García-Escudero et al. (2008).
Usage
data(M5data)
Format
The first two columns are the two variables. The last column is the true classification vector where symbol "0" stands for the contaminating data points.
Source
García-Escudero, L.A.; Gordaliza, A.; Matrán, C. and Mayo-Iscar, A. (2008), "A General Trimming Approach to Robust Cluster Analysis". Annals of Statistics, Vol.36, pp. 1324-1345.
Examples
#--- EXAMPLE 1 ------------------------------------------
data (M5data)
x <- M5data[, 1:2]
clus <- tclust(x, k=3, alpha=0.1, nstart=200, niter1=3, niter2=17,
nkeep=10, opt="HARD", equal.weights=FALSE, restr.fact=50, trace=TRUE)
plot (x, col=clus$cluster+1)