Premium
A model of the decision‐support scheme for express pest risk analysis using a Bayesian network implemented in Excel
Author(s) -
Holt J.,
Leach A. W.,
Schrader G.
Publication year - 2018
Publication title -
eppo bulletin
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.327
H-Index - 36
eISSN - 1365-2338
pISSN - 0250-8052
DOI - 10.1111/epp.12470
Subject(s) - weighting , judgement , risk assessment , variance (accounting) , computer science , bayesian network , statistics , uncertainty analysis , bayesian probability , risk analysis (engineering) , econometrics , actuarial science , operations research , mathematics , machine learning , artificial intelligence , business , medicine , accounting , computer security , political science , law , radiology
A model was developed to accompany the EPPO decision support scheme for express pest risk analysis (PRA) and provide a calculated overall risk and uncertainty for the PRA and so act as a reference for the judgement of overall risk and uncertainty provided by expert working groups. Implemented in Excel, it is readily accessible to PRA practitioners and offers: (a) a consistent and explained weighting of the different risk factors and a rationale for the way they are combined, (b) a calculated integration of the risk factor distributions to facilitate judgement of overall uncertainty, and (c) an account of the interaction between the rating and the uncertainty score so that, for example, an overall rating of moderate is not necessarily used to reflect uncertainty about assessments in which the risk is neither obviously high nor low. Of the nine published express PRA s examined, the rating and uncertainty predicted by the model were: in agreement with five; differed in both rating and uncertainty in one case; differed in rating only in one case; and differed in uncertainty only in two cases. Possible reasons for these differences were examined and the interpretation of model results to inform assessments is discussed.