Back to Search Start Over

Optimal designs for experiments for scalar-on-function linear models

Authors :
Michaelides, Damianos
Adamou, Maria
Woods, David C.
Overstall, Antony M.
Publication Year :
2021

Abstract

The aim of this work is to extend the usual optimal experimental design paradigm to experiments where the settings of one or more factors are functions. Such factors are known as profile factors, or as dynamic factors. For these new experiments, a design consists of combinations of functions for each run of the experiment. After briefly introducing the class of profile factors, basis functions are described with primary focus given on the B-spline basis system, due to its computational efficiency and useful properties. Basis function expansions are applied to a functional linear model consisting of profile factors, reducing the problem to an optimisation of basis coefficients. The methodology developed comprises special cases, including combinations of profile and non-functional factors, interactions, and polynomial effects. The method is finally applied to an experimental design problem in a Biopharmaceutical study that is performed using the Ambr250 modular bioreactor.

Subjects

Subjects :
Statistics - Methodology

Details

Database :
arXiv
Publication Type :
Report
Accession number :
edsarx.2110.09115
Document Type :
Working Paper