Aggregation Error in Bayesian Analysis of Reliability Systems

M. Naceur Azaiez and Vicki M. Bier
Additional contact information
M. Naceur Azaiez: Department of Industrial Engineering, University of Wisconsin-Madison, 1513 University Avenue, Madison, Wisconsin 53706
Vicki M. Bier: Department of Industrial Engineering, University of Wisconsin-Madison, 1513 University Avenue, Madison, Wisconsin 53706

Management Science, 1996, vol. 42, issue 4, 516-528

Abstract: Perfect aggregation in Bayesian system reliability analysis has been shown to be extremely unlikely. In other words, aggregation error is almost inevitable. Consequently, analysts have to deal with the following dilemma: on one hand, an aggregate analysis (i.e., an analysis at the system level), while relatively inexpensive, may be misleading. On the other hand, a disaggregate analysis (i.e., at the component level) provides more accurate results, but may be costly and impractical. Therefore, simple techniques to estimate the size of aggregation error are necessary to help analysts choose the most appropriate level of detail for an analysis. In this paper, reasonable bounds on the aggregation error are derived for a variety of reliability models. In particular, these bounds will never be more than twice the actual error. Tools to compute these bounds (and in some cases the actual error) are also provided.

Keywords: Bayesian estimation; perfect aggregation; aggregation error (search for similar items in EconPapers)
Date: 1996
References: Add references at CitEc
Citations: View citations in EconPapers (2)

Downloads: (external link)
http://dx.doi.org/10.1287/mnsc.42.4.516 (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:inm:ormnsc:v:42:y:1996:i:4:p:516-528

Access Statistics for this article

More articles in Management Science from INFORMS Contact information at EDIRC.
Bibliographic data for series maintained by Chris Asher ().