The Benefits of Accounting for Dependency in Empirical Bayes Estimation

Activity: Talk or presentation typesOral presentation

Description

For engineering problems in reliability management programs the use of all available information is very tempting. Especially while encountering a lack of data. However, such an approach must be handled with caution, because of encountering and being subject to possible fallacies. A simulation setup for situations, where results cannot be obtained analytically, provides us with a desired framework. The work presents the benefits of using appropriate expert engineering judgement. Our wish is to take into account underlying dependency between the data generating processes. By using a modified multivariate Poisson-Gamma model we are able to apprehend aleatory and epistemic uncertainty. For applications we are dependent on data, while the true underlying parameters of these inputs remain unknown, we need to rely on methods to estimate these parameters. Therefore, awareness of possible anticipated errors plays a central role.
Period17 Sep 2021
Event titleManagement Science Research Seminar
Event typeConference
LocationGlasgow, United Kingdom
Degree of RecognitionLocal