Example: model comparison

This example shows the multi-model comparison. You will see how to perform a multi-model comparison

Provided are three models, a linear models with 2 input parameters, a nonlinear model with 2 input parameters and a nonlinear model with 4 input parameters. The data to base the comparison on is given in an extra file.

Note

A detailed explanation of this example will be provided in future as part of the tutorial.

Model 1: L2_model

Pylink model1

Property	Setting
Model type	Function (linear)
Number of input parameters	2
Number of output parameters	1
Time- or space- dependency	space-dependency
MC reference	No

Priors1

Parameter	Distribution
0-2	given as correlated samples

Model 1: NL2_model

Pylink model1

Property	Setting
Model type	Function (exponential)
Number of input parameters	2
Number of output parameters	1
Time- or space- dependency	space-dependency
MC reference	No

Priors1

Parameter	Distribution
0-2	given as correlated samples

Model 1: NL4_model

Pylink model1

Property	Setting
Model type	Function (cosine)
Number of input parameters	4
Number of output parameters	1
Time- or space- dependency	space-dependency
MC reference	No

Priors1

Parameter	Distribution
0-4	given as correlated samples

Surrogates 1-3

All surrogates share the same setup and only differ in the given model.

MetaModel settings

Property	Setting
surrogate-type	aPCE
associated model	see lists above
degree choices	1-12, q-norm truncation 1.0
regression	OMP (Orthogonal matching pursuit)

Training choices

Property	Setting
Static sampling method	latin-hypercube
Number of static samples	100
Number of total samples	100