Probabilistic inversion of expert judgments in the quantification of model uncertainty

T.J. Bedford, B. Kraan

Research output: Contribution to journalArticle

28 Citations (Scopus)

Abstract

Expert judgment is frequently used to assess parameter values of quantitative management science models, particularly in decision-making contexts. Experts can, however, only be expected to assess observable quantities, not abstract model parameters. This means that we need a method for translating expert assessed uncertainties on model outputs into uncertainties on model parameter values. This process is called probabilistic inversion. The probability distribution on model parameters obtained in this way can be used in a variety of ways, but in particular in an uncertainty analysis or as a Bayes prior. This paper discusses computational algorithms that have proven successful in various projects and gives examples from environmental modelling and banking. Those algorithms are given a theoretical basis by adopting a minimum information approach to modelling partial information. The role of minimum information is two-fold: It enables us to resolve the problem of nonuniqueness of distributions given the information we have, and it provides numerical stability to the algorithm by guaranteeing convergence properties.
LanguageEnglish
Pages995-1006
Number of pages11
JournalManagement Science
Volume51
Issue number6
DOIs
Publication statusPublished - 2005

Fingerprint

Quantification
Expert judgment
Model uncertainty
Modeling
Uncertainty
Probability distribution
Banking
Management science
Uncertainty analysis
Decision making
Partial information

Keywords

  • multivariate distribution
  • uncertainty analysis
  • expert judgment
  • probabilistic inversion
  • credit scoring
  • environmental modeling

Cite this

@article{96d21d8e1a4e4b4b9a2b7771113d819a,
title = "Probabilistic inversion of expert judgments in the quantification of model uncertainty",
abstract = "Expert judgment is frequently used to assess parameter values of quantitative management science models, particularly in decision-making contexts. Experts can, however, only be expected to assess observable quantities, not abstract model parameters. This means that we need a method for translating expert assessed uncertainties on model outputs into uncertainties on model parameter values. This process is called probabilistic inversion. The probability distribution on model parameters obtained in this way can be used in a variety of ways, but in particular in an uncertainty analysis or as a Bayes prior. This paper discusses computational algorithms that have proven successful in various projects and gives examples from environmental modelling and banking. Those algorithms are given a theoretical basis by adopting a minimum information approach to modelling partial information. The role of minimum information is two-fold: It enables us to resolve the problem of nonuniqueness of distributions given the information we have, and it provides numerical stability to the algorithm by guaranteeing convergence properties.",
keywords = "multivariate distribution, uncertainty analysis, expert judgment, probabilistic inversion, credit scoring, environmental modeling",
author = "T.J. Bedford and B. Kraan",
year = "2005",
doi = "10.1287/mnsc.1050.0370",
language = "English",
volume = "51",
pages = "995--1006",
journal = "Management Science",
issn = "0025-1909",
number = "6",

}

Probabilistic inversion of expert judgments in the quantification of model uncertainty. / Bedford, T.J.; Kraan, B.

In: Management Science, Vol. 51, No. 6, 2005, p. 995-1006.

Research output: Contribution to journalArticle

TY - JOUR

T1 - Probabilistic inversion of expert judgments in the quantification of model uncertainty

AU - Bedford, T.J.

AU - Kraan, B.

PY - 2005

Y1 - 2005

N2 - Expert judgment is frequently used to assess parameter values of quantitative management science models, particularly in decision-making contexts. Experts can, however, only be expected to assess observable quantities, not abstract model parameters. This means that we need a method for translating expert assessed uncertainties on model outputs into uncertainties on model parameter values. This process is called probabilistic inversion. The probability distribution on model parameters obtained in this way can be used in a variety of ways, but in particular in an uncertainty analysis or as a Bayes prior. This paper discusses computational algorithms that have proven successful in various projects and gives examples from environmental modelling and banking. Those algorithms are given a theoretical basis by adopting a minimum information approach to modelling partial information. The role of minimum information is two-fold: It enables us to resolve the problem of nonuniqueness of distributions given the information we have, and it provides numerical stability to the algorithm by guaranteeing convergence properties.

AB - Expert judgment is frequently used to assess parameter values of quantitative management science models, particularly in decision-making contexts. Experts can, however, only be expected to assess observable quantities, not abstract model parameters. This means that we need a method for translating expert assessed uncertainties on model outputs into uncertainties on model parameter values. This process is called probabilistic inversion. The probability distribution on model parameters obtained in this way can be used in a variety of ways, but in particular in an uncertainty analysis or as a Bayes prior. This paper discusses computational algorithms that have proven successful in various projects and gives examples from environmental modelling and banking. Those algorithms are given a theoretical basis by adopting a minimum information approach to modelling partial information. The role of minimum information is two-fold: It enables us to resolve the problem of nonuniqueness of distributions given the information we have, and it provides numerical stability to the algorithm by guaranteeing convergence properties.

KW - multivariate distribution

KW - uncertainty analysis

KW - expert judgment

KW - probabilistic inversion

KW - credit scoring

KW - environmental modeling

UR - http://dx.doi.org/10.1287/mnsc.1050.0370

U2 - 10.1287/mnsc.1050.0370

DO - 10.1287/mnsc.1050.0370

M3 - Article

VL - 51

SP - 995

EP - 1006

JO - Management Science

T2 - Management Science

JF - Management Science

SN - 0025-1909

IS - 6

ER -