SelvarMix (version 1.2.1)

SelvarMix-package: Regularization for variable selection in model-based clustering and discriminant analysis

Description

SelvarMix is a package where a regularization approach of variable selection is considered in model-based clustering and discriminant analysis frameworks. First, this procedure consists of ranking the variables with a lasso-like procedure. Second, the method of Maugis et al (2009, 2011) is adapted to define the role of variables in the two frameworks. SelvarMix provides a faster variable selection algorithm than the backward stepwise or forward stepwise algorithms of Maugis et al (2009), allowing us to study high-dimensional datasets.

Details

Package:	SelvarMix
Type:	Package
Version:	1.0
Date:	2014-04-03
License:	GPL-3 + file LICENSE
LazyLoad:	yes

The general purpose of the package is to perform variable selection in model-based clustering and discriminant analysis. It focus on model-based clustering, where the clusters are assumed to arise from Gaussian distributions. The most achieved model in model-based clustering has been proposed by Maugis et al (2009). This so-called \(SRUW\) modeling considers three roles of variables: one variable may belong to the relevant clustering set \(S\), the redundant variable set \(U\) or the independent variable set \(W\). Moreover, the redundant variables may be explained by a subset \(R\) of the relevant variables \(S\). In order to avoid the greedy algorithms when high-dimensional data are studied, the SelvarMix procedure is proposed. It proceeds in two steps: First, the variables are ranked using a lasso-like procedure analogous to the one of Zhou et al (2009); second, the \(SRUW\) procedure is run on this ranked set of variables.

References

Maugis, C., Celeux, G., and Martin-Magniette, M. L., 2009. "Variable selection in model-based clustering: A general variable role modeling". Computational Statistics and Data Analysis, vol. 53/11, pp. 3872-3882.

Maugis, C., Celeux, G., and Martin-Magniette, M. L., 2011. "Variable selection in model-based discriminant analysis". Journal of Multivariate Analysis, vol. 102, pp. 1374-1387.

Zhou, H., Pan, W., and Shen, X., 2009. "Penalized model-based clustering with unconstrained covariance matrices". Electronic Journal of Statistics, vol. 3, pp.1473-1496.

Sedki, M., Celeux, G., Maugis-Rabusseau, C., 2014. "SelvarMix: A R package for variable selection in model-based clustering and discriminant analysis with a regularization approach". Inria Research Report available at http://hal.inria.fr/hal-01053784

Examples

# NOT RUN {
## wine data set 
## n = 178 observations, p = 27 variables 
data(wine)
## variable selection in model-based clustering  
set.seed(123)
obj <- SelvarClustLasso(x=wine[,1:27], nbcluster=1:5, nbcores=4)
summary(obj)
print(obj)

## variables selection in discriminant analysis
set.seed(123)
a <- seq(1, 178, 10)
b <- setdiff(1:178, a)
obj <- SelvarLearnLasso(x=wine[b,1:27], z=wine[b,28], xt=wine[a,1:27], zt=wine[a,28], nbcores=4)
summary(obj)
print(obj)
# }