multivar: A function to decompose the output data set and reduce its...

In multisensi: Multivariate Sensitivity Analysis

Description

The function multivar applies a multivariate method to decompose the output variables on a given basis.

Usage

multivar(simuls, dimension = NULL, reduction, centered = TRUE, 
         scale = TRUE, basis.args = list())

Arguments

simuls	a data.frame of size N x T, typically a set of N simulation outputs of length T
dimension	the number of variables to analyse, specified by an integer (for example 3) or by the minimal proportion of inertia (for example 0.95) to keep in the output decomposition
reduction	a function to decompose the multivariate output on a basis of smaller dimension
centered	logical value. If TRUE the output variables are centred.
scale	logical value. If TRUE the output variables are normalized.
basis.args	a list of arguments for the function given in the reduction argument. See the function help for more precision.

Value

A list containing:

H	a data.frame of size N x d, where d is the number of basis vectors. It contains the coefficients of the decomposition for each row of the simuls data.frame.
L	a matrix of size T x d. It contains the vectors of the user-defined basis.
sdev	standard deviations of the columns of H
nbcomp	number of components kept from the decomposition
SStot	total sums of squares of the simulations (after application of centered and scale)
centering	either 0 or the column averages of simuls
scaling	either 1 or sdY, depending on the scale argument
sdY	standard deviations of the columns of simuls
cor	correlation matrix (L*sdev), of size T x nbcomp
scale	kept in case the option scale has been changed in the function
importance	cumulated percentage of SS_H (sdev^2) with respect to SStot
call.info	list with the element reduction storing the name of the argument reduction

See Also

basis.ACP, basis.bsplines, basis.poly, basis.osplines

Examples

data(biomasseY)

res <- multivar(biomasseY, dimension=0.95, reduction=basis.ACP)

multisensi documentation built on May 2, 2019, 2:14 p.m.