Model Evaluation and Selection in Multiple Nonlinear Regression Analysis

Mathematical Modelling and Analysis 2007
Gints Jēkabsons, Jurijs Lavendels, Vjačeslavs Šitikovs

The main problem in regression model selection independently from application domain is finding the best model that best fits the data and does not neither overfit nor underfit. The aim of this work – to show one of possible ways to acquire adequate nonlinear regression models (parametric) of technical systems based on heuristic search and analytical optimality evaluation approach by taking into consideration the computational power of modern computers.

Atslēgas vārdi
Regression, approximation, model selection, heuristic search, model evaluation
DOI
10.3846/1392-6292.2007.12.81-90
Hipersaite
http://www.tandfonline.com/doi/abs/10.3846/1392-6292.2007.12.81-90#.VFEH3Ldxm1s

Jēkabsons, G., Lavendels, J., Šitikovs, V. Model Evaluation and Selection in Multiple Nonlinear Regression Analysis. Mathematical Modelling and Analysis, 2007, Vol.12, No.1, 81.-90.lpp. ISSN 1392-6292. e-ISSN 1648-3510. Pieejams: doi:10.3846/1392-6292.2007.12.81-90

Publikācijas valoda
English (en)