Multivariate elicitation: association, copulae, and graphical models

Alireza Daneshkhah, Tim Bedford

Research output: Chapter in Book/Report/Conference proceedingChapter

Abstract

In practical elicitation problems, we very often wish to elicit from the expert her knowledge about more than one uncertain quantity. When these are considered independent, the uncertainty about them is characterized simply by their marginal distributions; so it is sufficient to elicit the expert's knowledge about the quantities separately. When the independence assumption between the uncertain quantities of interest is not reasonable or has to be explored, the elicitation process becomes more complex. This is where we need to elicit information about the association between the variables in some way. Unfortunately, despite the growing literature about elicitation, this is an area where there is very little guidance to be found. In this article, we present some methods proposed in the literature to specify a multivariate distribution. We focus on the methods that use copulae and vines to construct a joint probability distribution. They enable us to build a multivariate distribution based on the elicited marginal distributions and their dependencies.
LanguageEnglish
Title of host publicationWiley Encyclopedia of Operations Research and Management Science
EditorsJames J Cochran, Louis Anthony Cox Jr, Pinar Keskinocak , Jeffrey P. Kharoufeh, J. Cole Smith
PublisherJohn Wiley & Sons Inc.
ISBN (Print)9780470400630
DOIs
Publication statusPublished - 29 Mar 2011

Fingerprint

Association Model
Copula Models
Elicitation
Graphical Models
Multivariate Distribution
Marginal Distribution
Copula
Joint Distribution
Guidance
Probability Distribution
Sufficient
Uncertainty
Knowledge

Keywords

  • copula
  • minimum-informative copula
  • vine
  • rank correlation

Cite this

Daneshkhah, A., & Bedford, T. (2011). Multivariate elicitation: association, copulae, and graphical models. In J. J. Cochran, L. A. Cox Jr, P. Keskinocak , J. P. Kharoufeh, & J. Cole Smith (Eds.), Wiley Encyclopedia of Operations Research and Management Science John Wiley & Sons Inc.. https://doi.org/DOI: 10.1002/9780470400531.eorms0290
Daneshkhah, Alireza ; Bedford, Tim. / Multivariate elicitation : association, copulae, and graphical models. Wiley Encyclopedia of Operations Research and Management Science. editor / James J Cochran ; Louis Anthony Cox Jr ; Pinar Keskinocak ; Jeffrey P. Kharoufeh ; J. Cole Smith . John Wiley & Sons Inc., 2011.
@inbook{69d238bb4ccf4dc19041de1abcd1f5cb,
title = "Multivariate elicitation: association, copulae, and graphical models",
abstract = "In practical elicitation problems, we very often wish to elicit from the expert her knowledge about more than one uncertain quantity. When these are considered independent, the uncertainty about them is characterized simply by their marginal distributions; so it is sufficient to elicit the expert's knowledge about the quantities separately. When the independence assumption between the uncertain quantities of interest is not reasonable or has to be explored, the elicitation process becomes more complex. This is where we need to elicit information about the association between the variables in some way. Unfortunately, despite the growing literature about elicitation, this is an area where there is very little guidance to be found. In this article, we present some methods proposed in the literature to specify a multivariate distribution. We focus on the methods that use copulae and vines to construct a joint probability distribution. They enable us to build a multivariate distribution based on the elicited marginal distributions and their dependencies.",
keywords = "copula, minimum-informative copula, vine, rank correlation",
author = "Alireza Daneshkhah and Tim Bedford",
year = "2011",
month = "3",
day = "29",
doi = "DOI: 10.1002/9780470400531.eorms0290",
language = "English",
isbn = "9780470400630",
editor = "Cochran, {James J} and {Cox Jr}, {Louis Anthony} and {Keskinocak }, {Pinar } and Kharoufeh, {Jeffrey P. } and {Cole Smith }, {J. }",
booktitle = "Wiley Encyclopedia of Operations Research and Management Science",
publisher = "John Wiley & Sons Inc.",

}

Daneshkhah, A & Bedford, T 2011, Multivariate elicitation: association, copulae, and graphical models. in JJ Cochran, LA Cox Jr, P Keskinocak , JP Kharoufeh & J Cole Smith (eds), Wiley Encyclopedia of Operations Research and Management Science. John Wiley & Sons Inc. https://doi.org/DOI: 10.1002/9780470400531.eorms0290

Multivariate elicitation : association, copulae, and graphical models. / Daneshkhah, Alireza; Bedford, Tim.

Wiley Encyclopedia of Operations Research and Management Science. ed. / James J Cochran; Louis Anthony Cox Jr; Pinar Keskinocak ; Jeffrey P. Kharoufeh; J. Cole Smith . John Wiley & Sons Inc., 2011.

Research output: Chapter in Book/Report/Conference proceedingChapter

TY - CHAP

T1 - Multivariate elicitation

T2 - association, copulae, and graphical models

AU - Daneshkhah, Alireza

AU - Bedford, Tim

PY - 2011/3/29

Y1 - 2011/3/29

N2 - In practical elicitation problems, we very often wish to elicit from the expert her knowledge about more than one uncertain quantity. When these are considered independent, the uncertainty about them is characterized simply by their marginal distributions; so it is sufficient to elicit the expert's knowledge about the quantities separately. When the independence assumption between the uncertain quantities of interest is not reasonable or has to be explored, the elicitation process becomes more complex. This is where we need to elicit information about the association between the variables in some way. Unfortunately, despite the growing literature about elicitation, this is an area where there is very little guidance to be found. In this article, we present some methods proposed in the literature to specify a multivariate distribution. We focus on the methods that use copulae and vines to construct a joint probability distribution. They enable us to build a multivariate distribution based on the elicited marginal distributions and their dependencies.

AB - In practical elicitation problems, we very often wish to elicit from the expert her knowledge about more than one uncertain quantity. When these are considered independent, the uncertainty about them is characterized simply by their marginal distributions; so it is sufficient to elicit the expert's knowledge about the quantities separately. When the independence assumption between the uncertain quantities of interest is not reasonable or has to be explored, the elicitation process becomes more complex. This is where we need to elicit information about the association between the variables in some way. Unfortunately, despite the growing literature about elicitation, this is an area where there is very little guidance to be found. In this article, we present some methods proposed in the literature to specify a multivariate distribution. We focus on the methods that use copulae and vines to construct a joint probability distribution. They enable us to build a multivariate distribution based on the elicited marginal distributions and their dependencies.

KW - copula

KW - minimum-informative copula

KW - vine

KW - rank correlation

U2 - DOI: 10.1002/9780470400531.eorms0290

DO - DOI: 10.1002/9780470400531.eorms0290

M3 - Chapter

SN - 9780470400630

BT - Wiley Encyclopedia of Operations Research and Management Science

A2 - Cochran, James J

A2 - Cox Jr, Louis Anthony

A2 - Keskinocak , Pinar

A2 - Kharoufeh, Jeffrey P.

A2 - Cole Smith , J.

PB - John Wiley & Sons Inc.

ER -

Daneshkhah A, Bedford T. Multivariate elicitation: association, copulae, and graphical models. In Cochran JJ, Cox Jr LA, Keskinocak P, Kharoufeh JP, Cole Smith J, editors, Wiley Encyclopedia of Operations Research and Management Science. John Wiley & Sons Inc. 2011 https://doi.org/DOI: 10.1002/9780470400531.eorms0290