Blind Kriging: Implementation and performance analysis I. Couckuyt a , A. Forrester b , D. Gorissen b , F. De Turck a , T. Dhaene a a Ghent University-IBBT, Dept. of Information Technology (INTEC), Gaston Crommenlaan 8, 9000 Ghent b University of Southampton, School of Engineering Sciences, University Road, Southampton, United Kingdom Abstract When analysing data from computationally expensive simulation codes or pro- cess measurements, surrogate modelling methods are firmly established as facil- itators for design space exploration, sensitivity analysis, visualisation and op- timisation. Kriging is a popular surrogate modelling technique for data based on deterministic computer experiments. There exist several types of Kriging, mostly differing in the type of regression function used. Recently a promising new variable selection technique was proposed to identify a regression function in the Kriging framework. In this paper this type of Kriging, i.e., blind Kri- ging, has been efficiently implemented in Matlab® and has been extended. The implementation is validated and tested on several examples to illustrate the strength and weaknesses of this new, promising modelling technique. It is shown that the performance of blind Kriging is as good as, or better than or- dinary Kriging. Though, blind Kriging comes at double the computational cost with respect to ordinary Kriging. Keywords: blind Kriging, surrogate modelling, feature selection, variable subset selection, benchmark 1. Introduction Many complex real world phenomena are difficult to study directly using controlled experiments. Instead, the use of computer simulations has become Email address: ivo.couckuyt@ugent.be (I. Couckuyt) Preprint submitted to Elsevier 6th February 2012