Title: Meta-heuristic to estimate parameters in Non-Linear Regression Models

Authors: K. Antony Arokia Durai Raj, B. Kanagasabapathi, Gopichand Agnihothram

Addresses: CKDIS, Software Engineering and Technology Labs, Infosys Technologies Limited, Bangalore 560 100, India. ' CKDIS, Software Engineering and Technology Labs, Infosys Technologies Limited, Bangalore 560 100, India. ' CKDIS, Software Engineering and Technology Labs, Infosys Technologies Limited, Bangalore 560 100, India

Abstract: Non-Linear Regression Models (NLRM) are used in analysing scientific applications such as metal treatment, chemical process, pharmacology, and physiology. If the parameters in a regression model are non-linear, then the model is termed as NLRM, even if the explanatory variables of such a model are linear. The computational effort required to solve linear regression models are less compared to NLRMs. In this paper we propose a Genetic Algorithm (GA) to estimate the parameters in NLRMs. The computational results show that the proposed GA performs better than/equivalent to the existing methods in most of the problem instances considered in this study.

Keywords: NLRM parameters; nonlinear regression models; parameter estimation; heuristics; GAs; genetic algorithms; modelling; metaheuristics.

DOI: 10.1504/IJMOR.2011.042439

International Journal of Mathematics in Operational Research, 2011 Vol.3 No.5, pp.473 - 489

Published online: 12 Feb 2015 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article