Utilisation de Donnees Intermediaires pour Corriger la Prediction de Modeles Mecanistes

A strategy is proposed for correcting the prediction of the final state of a system by a deterministic mechanistic model, based on intermediate observations of the system. The number and dates of the observations are considered to be random. The proposed correction term is based on the best linear u...

Full description

Saved in:
Bibliographic Details
Published in:Biometrics 1991-03, Vol.47 (1), p.1-12
Main Authors: Faivre, R., Goffinet, B., Wallach, D.
Format: Article
Language:eng ; fre
Citations: Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:A strategy is proposed for correcting the prediction of the final state of a system by a deterministic mechanistic model, based on intermediate observations of the system. The number and dates of the observations are considered to be random. The proposed correction term is based on the best linear unbiased predictor of the error in final state prediction, based on the errors of prediction for the observed variables. Assumptions are made that make it possible to estimate the terms in this predictor, for any new situation, from a training set. On an example of wheat stand, the use of intermediate measurements of biomass reduced the mean squared error of prediction (MSEP) of yield by 80%.
ISSN:0006-341X
1541-0420
DOI:10.2307/2532490