pawn_generic: Computation of PAWN indices (generic approach)

In arnaldpuy/pawnr: Computation of PAWN Sensitivity Indices

Description

This function computes and bootstraps PAWN indices following the "generic approach" of \insertCitePianosi2018;textualpawnr.

Usage

pawn_generic(data, Y, n, test, R)

Arguments

data	A data frame or matrix with the sample matrix.
Y	Numeric vector, model output.
n	Integer, number of conditioning intervals.
test	Summary statistic (i.e. mean, median, max).
R	Integer, number of bootstrap replicas.

Details

The bootstrap is conducted without replacement, as recommended by \insertCiteKhorashadiZadeh2017;textualpawnr.

Value

A data table with the bootstrap PAWN indices of each model input.

References

\insertAllCited

Examples

# Create matrix:
A <- randtoolbox::sobol(n = 100, dim = 8)
# Compute model output:
Y <- sobol_Fun(A)
T <- pawn_generic(data = A, Y = Y, n = 10, test = median, R = 50 )

arnaldpuy/pawnr documentation built on Sept. 20, 2019, 12:15 p.m.