ASSESSMENT OF CONDENSED MODELS FOR THREE-FACTOR BOX-BEHNKEN DESIGNS FOR HYPERSPHERE RESPONSE SURFACE INVESTIGATION
Advances and Applications in Statistics · 06.08.2026 · 93 (8) · 945-962
DOI: 10.17654/0972361726045 ↗Kurz gefasst
Fachartikel in der Zeitschrift Advances and Applications in Statistics, online erschienen am 06.08.2026. Verfasst von T. A. Ugbe und 7 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Box-Behnken designs are very effective second-order response surface designs that are frequently utilized for optimization in both industry and science. However, for most systems, the entire quadratic model may be over-parameterized, resulting in estimation and prediction inefficiencies. In order to determine the best models for response surface exploration within a spherical design region (hypersphere), this study critically assesses 43 reduced models that were developed from a three-factor Box-Behnken design. To evaluate prediction variance performance, we used fraction of design space graphs in conjunction with established design optimality criteria, such as $D$-,$A$-,$G$-, and $V$-efficiency. The findings repeatedly show that the full quadratic model is significantly outperformed by reduced models that do not include quadratic terms. The most straightforward model with just one linear term (Design 43, $p=2)$ produced the highest $D$- (70.71), $A$- (66.67), and $V$- (2.2016) efficiencies. The model with the highest $G$-efficiency (57.14) has all three linear terms and no quadratic or interaction terms (Design 34, $p=4)$. Models without quadratic components show better and more consistent prediction variance throughout the whole design space, according to fraction of design space analysis. The results significantly support strategic model reduction, especially the elimination of quadratic terms, to improve Box-Behnken design performance for robust prediction and effective parameter estimation in response surface approach.
Zitieren
APA
Ugbe, T. A., Okon, E. A., Akpan, S. S., Macgeorge, C. G., Ofem, D. I., Owan, R. A., Dooga, M. Y., & Ayobami, S. A. (2026). ASSESSMENT OF CONDENSED MODELS FOR THREE-FACTOR BOX-BEHNKEN DESIGNS FOR HYPERSPHERE RESPONSE SURFACE INVESTIGATION. Advances and Applications in Statistics, 93(8), 945-962. https://doi.org/10.17654/0972361726045
Vancouver
Ugbe TA, Okon EA, Akpan SS, Macgeorge CG, Ofem DI, Owan RA, et al. ASSESSMENT OF CONDENSED MODELS FOR THREE-FACTOR BOX-BEHNKEN DESIGNS FOR HYPERSPHERE RESPONSE SURFACE INVESTIGATION. Advances and Applications in Statistics. 2026;93(8):945-962. doi:10.17654/0972361726045
BibTeX
@article{Ugbe2026,
author = {Ugbe, T. A. and Okon, E. A. and Akpan, S. S. and Macgeorge, C. G. and Ofem, D. I. and Owan, R. A. and Dooga, M. Y. and Ayobami, S. A.},
title = {ASSESSMENT OF CONDENSED MODELS FOR THREE-FACTOR BOX-BEHNKEN DESIGNS FOR HYPERSPHERE RESPONSE SURFACE INVESTIGATION},
journal = {Advances and Applications in Statistics},
year = {2026},
volume = {93},
number = {8},
pages = {945--962},
doi = {10.17654/0972361726045}
}
EUVIMED