Modalclust (version 0.7)

phmac: Main function for performing Modal Clusters either parallel or serial mode.

Description

Performing Modal Clustering

Usage

phmac(dat, length = 10, npart = 1, parallel = TRUE, sigmaselect = NULL,
G= NULL)
modalclust(dat, length = 10, npart = 1, parallel = TRUE, sigmaselect = NULL,
G= NULL)

Arguments

dat

Matrix of data points

length

number of smoothing levels. Default is 10

sigmaselect

Specified Smoothing levels. Default NULL will calculate the Sigma levels using concept of spectral degrees of freedom given in Lindsay et al (2008)

npart

Number of random partitions when using parallel computing. If using several processors of a machine one option is to choose the number of partitions equal to the number of processors

parallel

If TRUE uses parallel comptation using npart processors. Requires the package multicore to perform parallel computing

G

Specified values of modes. A matrix with number or rows equal to the number of modes and number of columns equal to the dimension of the data. Defualt value is NULL

Value

data

Same as the input Data

n.cluster

Number of clusters at each level.

level

Levels corresponding to each smoothing parameter.

sigmas

Same as input sigmaselect if provided or dynamically calculated smoothing levels based on Spectral Degrees of Freedom criterion. Uses the function khat.inv

mode

List of modes at each distinct levels.

membership

List of memmbership to modes at each distinct levels.

References

Li. J, Ray. S, Lindsay. B. G, "A nonparametric statistical approach to clustering via mode identification," Journal of Machine Learning Research , 8(8):1687-1723, 2007.

Lindsay, B.G., Markatou M., Ray, S., Yang, K., Chen, S.C. "Quadratic distances on probabilities: the foundations," The Annals of Statistics Vol. 36, No. 2, page 983--1006, 2008.

See Also

soft.hmac for soft clustering at specified levels. hard.hmac for hard clustering at specified levels. See plot.hmac.

Examples

# NOT RUN {
data(disc2d)
# }
# NOT RUN {
disc2d.hmac=phmac(disc2d,npart=1)
# }
# NOT RUN {
plot.hmac(disc2d.hmac,level=2)


## For parallel implementation
# }
# NOT RUN {
disc2d.hmac.parallel=phmac(disc2d,npart=2,parallel=TRUE)
# }
# NOT RUN {
soft.hmac(disc2d.hmac,level=2)
soft.hmac(disc2d.hmac,n.cluster=3)

hard.hmac(disc2d.hmac,n.cluster=3)
# }