Premium
OPTIMAL EXPERIMENT PLANNING FOR ENVIRONMENTAL KINETIC MODELS SUBJECT TO DATA HETEROSCEDASTICITY
Author(s) -
DOVI' VINCENZO G.
Publication year - 1997
Publication title -
environmetrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.68
H-Index - 58
eISSN - 1099-095X
pISSN - 1180-4009
DOI - 10.1002/(sici)1099-095x(199707)8:4<303::aid-env252>3.0.co;2-z
Subject(s) - heteroscedasticity , computer science , exponential function , constant (computer programming) , statistics , econometrics , mathematical optimization , mathematics , mathematical analysis , programming language
Accurate design of sequential experiments is fundamental for the precise estimation of parameters in environmental kinetic models, due to the high cost of each single experiment. At each step the information available must be used effectively so that each new experiment provides the largest amount of additional information. The most severe difficulty in the determination of optimal settings is the high correlation between pre‐exponential factors and other parameters. It has been demonstrated in a previous paper that, in the cases of constant absolute and relative errors, this difficulty can be overcome by the method of parameter separation. In this paper we show that the method can be successfully employed also in the case of a general heteroscedasticity structure of the experimental errors.© 1997 John Wiley & Sons, Ltd.