kriging.model.lsd {LSDsensitivity}

Fit a Kriging meta-model to a LSD model sample data

Description

This function fits a Kriging meta-model (also known as a Gaussian process), using five alternative variance kernels and two trend model options, to the sampled data from a LSD simulation model.

Usage

kriging.model.lsd( data, ext.wgth = 0.5, trendModel = 0, covModel = 0,
                   digits = 4 )

Arguments

data

an object created by a previous call to read.doe.lsd which contains all the required experimental data for the analysis.

ext.wgth

numeric in [0, 1]: the weight given to the fitting metrics calculated over the out-of-sample (external) validation sample in regard to the in-sample metrics. The default value is 0.5.

trendModel

a number corresponding to the trend model: 0 = automatic selection (according to fitting metrics, the default); 1 = constant; 2 = first order polynomial.

covModel

a number corresponding to the covariance model (or kernel): 0 = automatic selection (according to fitting metrics, the default); 1 = Matern 5/2; 2 = Matern 3/2; 3 = Gaussian; 4 = exponential; 5 = power exponential.

digits

integer: the number of significant digits to show in results. The default is 4.

Details

This function fits a universal Kriging meta-model to the experimental data set previously loaded with read.doe.lsd using the Gaussian process method (Rasmussen & Williams, 2006).

This function is a wrapper to the function km in DiceKriging-package.

Value

The function returns an object/list of class kriging-model containing several items:

selected

an object containing the selected estimated meta-model (standardized).

comparison

a print-ready table with all fitting statistics for all fitted meta-model specifications.

Q2

the Q2 in-sample fitting statistic for the selected meta-model.

rmse

the RMSE out-of-sample fitting statistic for the selected meta-model.

mae

the MAE out-of-sample fitting statistic for the selected meta-model.

rma

the RMA out-of-sample fitting statistic for the selected meta-model.

extN

number of out-of-sample observations.

estimation

a print-ready table with the coefficients (hyper-parameters) of the selected estimated meta-model.

estimation.std

a print-ready table with the standardized coefficients (hyper-parameters) of the selected estimated meta-model.

coefficients

a vector with the coefficients (hyper-parameters) of the selected estimated meta-model.

coefficients.std

a vector with the standardized coefficients (hyper-parameters) of the selected estimated meta-model.

trend

number of the selected trend model.

trendNames

name of the selected trend model.

cov

number of the selected covariance model (kernel).

covNames

name of the selected covariance model (kernel).

Note

See the note in LSDsensitivity-package for step-by-step instructions on how to perform the complete sensitivity analysis process using LSD and R.

Author(s)

NA

References

Kleijnen JP (2009) Kriging metamodeling in simulation: a review. Eur J Oper Res 192(3):707-716

Rasmussen C, Williams C (2006) Gaussian processes for machine learning. MIT Press, Cambridge

Roustant O, Ginsbourger D, Deville Y (2012) Dicekriging, diceoptim: two R packages for the analysis of computer experiments by kriging-based metamodeling and optimization. J Stat Softw 51(1):1-55

See Also

read.doe.lsd()

km in DiceKriging-package

Examples

# get the example directory name
path <- system.file( "extdata/sobol", package = "LSDsensitivity" )

# Steps to use this function:
# 1. define the variables you want to use in the analysis
# 2. load data from a LSD simulation saved results using read.doe.lsd,
#    preferrably using two sets of sampled data (DoEs), one for model
#    estimation and the other for out-of-sample (external) validation
# 3. fit a Kriging (or polynomial) meta-model using kriging.model.lsd

lsdVars <- c( "var1", "var2", "var3" )         # the definition of existing variables

dataSet <- read.doe.lsd( path,                 # data files folder
                         "Sim3",               # data files base name (same as .lsd file)
                         "var3",               # variable name to perform the sensitivity analysis
                         does = 2,             # number of experiments (data + external validation)
                         saveVars = lsdVars )  # LSD variables to keep in dataset

model <- kriging.model.lsd( dataSet )          # estimate best Kriging meta-model

print( model$comparison )                      # model comparison table
print( model$estimation.std )                  # model estimation (standardized) table