dprep (version 3.0.2)

lvf: Las Vegas Filter

Description

Las Vegas Filter uses a random generation of subsets and an inconsistency measure as the evaluation function to determine the relevance of features in the dataset.

Usage

lvf(data, lambda, maxiter)

Arguments

data

Name of the discretized dataset

lambda

Threshold for the inconsistency

maxiter

Maximum number of iterations

Value

bestsubset

The best subset of features

Details

If the dataset has continuous variables, these must first be discretized. This package includes four discretization methods. A value of lambda close to the inconsistency of the whole dataset yields a large number of selected features, a large lambda yields few selected features.

References

LIU, H. and SETIONO, R. (1996). A probabilistic approach to feature selection: a filter solution. Proc. of the thirteenth International Conference of Machine Learning, 319-337.

See Also

disc.ew,inconsist,finco

Examples

#---- LVF method ----
data(iris)
iris.discew=disc.ew(iris,1:4,out="num")
inconsist(iris.discew)
lvf(iris.discew,0,100)