hypervolume_variable_importance: Hypervolume variable importance

In dc165/Hypervolume-Dev: High Dimensional Geometry and Set Operations Using Kernel Density Estimation, Support Vector Machines, and Convex Hulls

Description

Assesses the contribution of each variable to the total hypervolume as a rough metric of variable importance.

Usage

hypervolume_variable_importance(hv, verbose = TRUE)

Arguments

hv	A hypervolume for which the importance of each variable should be calculated.
verbose	If TRUE, prints diagnostic progress messages.

Details

The algorithm proceeds by comparing the n-dimensional input hypervolume's volume to all possible n-1 dimensional hypervolumes where each variable of interest has been deleted. The importance score reported is the ratio of the n-dimensional hypervolume relative to each of the n-1 dimensional hypervolumes. Larger values indicate that a variable makes a proportionally higher contribution to the overall volume.

The algorithm can only be used on Hypervolumes that have a Data and Method value, because the variable deletion process is not well defined for objects that are not associated with a particular set of observations and construction method.

Value

A named vector with importance scores for each axis. Note that these scores are not dimensionless but rather have units corresponding to the original units of each variable.

Examples

# low parameter values for speed
data(iris)
hv = hypervolume_gaussian(subset(iris, Species=="versicolor")[,1:2],samples.per.point=10)
varimp = hypervolume_variable_importance(hv,verbose=FALSE)
barplot(varimp)

dc165/Hypervolume-Dev documentation built on Dec. 13, 2020, 6:02 p.m.