Classifies data into two groups using a clustering regime.

Description

This function is primarily designed to split a bimodal data vector into two groups to allow the fitting of mixture generalised lambda distributions.

Usage

1
fun.class.regime.bi(data, perc.cross, fun.cross)

Arguments

data

Data to be classified into two groups.

perc.cross

Percentage of cross over from one data to the other, usually set at 1%

fun.cross

Any clustering function such as link{clara}, pam, fanny can be used here. Or a logical vector indicating how data should be split.

Details

This function is part of the routine mixture fitting procedure provided in this package. The perc.cross argument or percentage of cross over is designed to allow the use of maximum likelihood estimation via EM algorithm for fitting bimodal data. When this is invoked, it will ensure both part of the data will contain both the minmum and maximum of the data set as well as a proportion ( specified in perc.cross argument) of observations from each other. If 1% is required, then data.a will contains 1% of the data.b and vice versa after the full data set has been classified into data.a and data.b by the fun.cross classification regime.

Value

data.a

First group of data obtained by the classification algorithm.

data.b

Second group of data obtained by the classification algorithm.

Author(s)

Steve Su

References

Kaufman, L. and Rousseeuw, P. J. (1990). Finding Groups in Data: An Introduction to Cluster Analysis. Wiley, New York.

Su (2006) Maximum Log Likelihood Estimation using EM Algorithm and Partition Maximum Log Likelihood Estimation for Mixtures of Generalized Lambda Distributions. Working Paper.

See Also

link{clara}, pam, fanny

Examples

1
2
3
4
5
## Classify the faithful[,1] data into two categories with 10% cross over mix.
# fun.class.regime.bi(faithful[,1],0.1,clara)

## Classify the faithful[,1] data into two categories with no mixing:
# fun.class.regime.bi(faithful[,1],0,clara)

Want to suggest features or report bugs for rdrr.io? Use the GitHub issue tracker.