normalize: Normalize approximations set(s).

In ecr: Evolutionary Computation in R

Normalize approximations set(s).

Description

Normalization is done by subtracting the min.value for each dimension and dividing by the difference max.value - min.value for each dimension Certain EMOA indicators require all elements to be strictly positive. Hence, an optional offset is added to each element which defaults to zero.

Usage

normalize(x, obj.cols, min.value = NULL, max.value = NULL, offset = NULL)

Arguments

x	[matrix | data.frame] Either a numeric matrix (each column corresponds to a point) or a data.frame with columns at least obj.cols.
obj.cols	[character(>= 2)] Column names of the objective functions.
min.value	[numeric] Vector of minimal values of length nrow(x). Only relevant if x is a matrix. Default is the row-wise minimum of x.
max.value	[numeric] Vector of maximal values of length nrow(x). Only relevant if x is a matrix. Default is the row-wise maximum of x.
offset	[numeric] Numeric constant added to each normalized element. Useful to make all objectives strictly positive, e.g., located in [1,2].

Value

[matrix | data.frame]

Note

In case a data.frame is passed and a “prob” column exists, normalization is performed for each unique element of the “prob” column independently (if existent).

See Also

Other EMOA performance assessment tools: approximateNadirPoint(), approximateRefPoints(), approximateRefSets(), computeDominanceRanking(), emoaIndEps(), makeEMOAIndicator(), niceCellFormater(), plotDistribution(), plotFront(), plotScatter2d(), plotScatter3d(), toLatex()

ecr documentation built on March 31, 2023, 10:07 p.m.