levelsetdist {bigdatadist}

Level Set Distances

Description

This function allows you to compute the alpha level set distances as proposed in Martos et al. Nomparametric distances for probability measures with applications, 2018 (submitted).

Usage

levelsetdist(A,B,n.level=10,k.neighbor=10) 

Arguments

A

data set in a matrix where variables are in columns and observations are in rows.

B

data set in a matrix where variables are in columns and observations are in rows.

n.level

the number of level sets to consider for distance computation.

k.neighbor

number of neighbour points to consider in the estimation of the support of the distribution on each class.

Details

The function computes the alpha level set distance between two (samples from) different probability measures. Details about the distance and the criterion to fix its hyperparameter can be found in Martos et al (2018, submitted).

Value

distance

distance estimation between the two data sets or distributions.

References

Martos, G. et al (2018): Nomparametric distances for probability measures with applications in classification. J. of Calssification, 2018 (submitted).

Examples

require(MASS);
set.seed(1)
A = mvrnorm(100,c(0,0),diag(2));  B = mvrnorm(150,c(1,1),diag(2)); 
levelsetdist(A, B)