This article identifies the basic elements of a decision model and the errors that can be made in the development of optimizing decision models–errors of specification and errors of exploration. Specification errors are defined as those errors that result from the use of inaccurate structural relationships and parameters, whereas exploration errors occur in the process of searching the decision space for the optimal response. A number of systematic search procedures that can be used to reduce the exploration error in simulation models are reviewed. These procedures include random search methods, experimental designs, “learning mechanisms,” and “hill climbing.” It is suggested that if the availability of these procedures is recognized, the choice of a particular empirical model can be based on its ability to accurately portray the decision problem rather than the availability of an error-free optimizing algorithm.
Ľarticle traile des principes caraclerisliques fondamentaux concernant les modeles de norme ainsi que des erreurs possibles dans la mise au point de modeles de norme a reglage optimum, surtout des erreurs dues a la specification et aľexploration. Les erreurs dites specif ques proviennent de ľemploi de rapports el de paramelres fonctionellement fauiifs ou inexacts, tandis que les erreurs ďexploration sont le resullat de calculs imprecis de ľespace denorme convenable pour obtenir une reponse optima. Ľauteur presente une serie ďexemples en vue de systematiser les procedes valables qui aboulissenl a une reduction des erreurs ďexploration aľaide de modeles-simulaieurs. Parmi les procedes convenables on propose une methode de determinations au hasard, certains projets experimentaux, des “mecanismes ďassimilalion intellectuelle” et des “experiences de tâtonnemenl progressif” L'auteur en conclut que si ľon reconnat et qu'on emploie les methodes proposees, il est bien possible de mettre au pooint un modele experimental ou empirique capable de servir de modele problemalique bien exact plutol que de s'appuyer foncieremenl sur ľemploi ďun algorithme optimum valable.
[1]
Robert C. Meier.
The Application of Optimum-Seeking Techniques to Simulation Studies: A Preliminary Evaluation
,
1967
.
[2]
H. Hotelling.
Experimental Determination of the Maximum of a Function
,
1941
.
[3]
C. Hildreth.
Problems of Uncertainty in Farm Planning
,
1957
.
[4]
J. S. Hunter,et al.
Multi-Factor Experimental Designs for Exploring Response Surfaces
,
1957
.
[5]
Wilfred Candler,et al.
Estimation of Performance Functions for Budgeting and Simulation Studies
,
1969
.
[6]
Samuel H. Brooks.
A Discussion of Random Methods for Seeking Maxima
,
1958
.
[7]
Earl O. Heady,et al.
Response Surface Analysis and Simulation Models in Policy Choices
,
1970
.
[8]
Samuel H. Brooks.
A Comparison of Maximum-Seeking Methods
,
1959
.
[9]
G. Box.
The Exploration and Exploitation of Response Surfaces: Some General Considerations and Examples
,
1954
.
[10]
Thomas H. Naylor,et al.
Experimental Designs for Computer Simulation Experiments
,
1970
.
[11]
Pinhas Zusman,et al.
Simulation: A Tool for Farm Planning under Conditions of Weather Uncertainty
,
1965
.