UP: UP: Universal Prediction Distribution

In malekbs/UP: What the package does (short line)

Description

UP is a package that provides universal method for surrogate models: A package with extensible options and various UP-based algorithms for optimization, refienment and inversion.

Introduction

Universal Prediction Distribution for surrogate models is an efficient too for adaptive designs. It allows the extension of many algorithms defined for the Gaussian case for all types of surrogates

UP Distribution

UP of a surogate model is given by the prediction of its submodels defined by a resampling technique.

Methods

The package also contains several algorihms that uses the UP distribution such as: upego, upsmart, upinverse and ga model selection.

Applications

UP applications are diverse. We can cite mainely: sequential sampling, optimization and model selection. UPis always useful when one uses surrogate modeling paradigm.

See Also

UP

Examples

library(UP)
x           <- as.matrix(c(-2.6,-0.2, 1.7,-1.4,1.2,3))
y           <- c(0.8, 0.5, 0.1, 0.3, 0, 0.4)
xverif      <- seq(-3, 3, length.out =300)
krig        <- krigingsm$new()
resampling  <- UPClass$new(x, y, Scale =TRUE) 
upsm        <- UPSM$new(sm= krig, UP= resampling) 
prediction  <- upsm$uppredict(xverif)
plotUP1D(xverif, prediction, x, y)

malekbs/UP documentation built on May 14, 2019, 8:05 a.m.