T O W A R D S  O P T I M A L  D E S I G N  O F N O N L I N E A R  R E G R E S S I O N  M O D E L S
Cieszewski CJ and I.E. Bella
Proceedings of a IUFRO S4.11 Conference held on 10-14 September 1991 at the University of Greenwich  

London, UK    

                                    
Abstract:

This paper presents ideas and examples of biological and mathematical means that can facilitate nonlinear regression model development. They are particularly important for development of models from scant data, where overall trends are not easily discernable or when one needs to extrapolate beyond available data. The ideas cover mathematical techniques for stabilizing equations and their coefficients for efficient model parameterization; shortcuts in regression analysis for reducing model dimensionality; and function generalizations for achieving a unified approach to model selection. Some application of these ideas are illustrated through a case study of a stand growth model development for managed, second growth lodgepole pine, based on natural unmanaged stand data. 

Author Keywords:
Nonlinear regression model, model derivation and stability, biological models, scant data.

KeyWords Plus:

Addresses:
Cieszewski CJ, Univ Georgia, Sch Forest Resources, Athens, GA 30602 USA
Univ Georgia, Sch Forest Resources, Athens, GA 30602 USA

Publisher:

IDS Number:

ISSN: