Premium
An approach to perform expert elicitation for engineering design risk analysis: methodology and experimental results
Author(s) -
Babuscia Alessandra,
Cheung KarMing
Publication year - 2014
Publication title -
journal of the royal statistical society: series a (statistics in society)
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.103
H-Index - 84
eISSN - 1467-985X
pISSN - 0964-1998
DOI - 10.1111/rssa.12028
Subject(s) - expert elicitation , weighting , computer science , calibration , quality (philosophy) , machine learning , design of experiments , expert system , aggregate (composite) , data mining , risk analysis (engineering) , artificial intelligence , statistics , mathematics , medicine , philosophy , materials science , epistemology , composite material , radiology
Summary Expert elicitation is increasingly applied to different research areas. Multiple approaches have been implemented, but the development of methods to quantify experts' biases and calibration represents a challenge. As a result, the integration of multiple and often conflicting opinions can be demanding, owing to the complexity of properly weighting experts' contributions. We propose an approach to address this problem when probability densities for seed calibration variables are not available. The methodology generates an expert score that is employed to aggregate multiple‐expert assessments. The approach has been experimentally applied to engineering design risk analysis. Results indicate that the approach improves the quality of the estimations. The weighted aggregations of experts' estimates based on the experts' scores achieve better results than the corresponding aggregations based on experts' opinions equally weighted.