UNCERTAINTY ANALYSIS WITH MACCS2 USING DATA BASED ON EXPERT ELICITATION

摘要

Results of uncertainty analyses are highly dependant on the values for input distributions. Caution should be used when accepting uncertainty distributions provided by others or from sources that are not fully understood. To help build confidence, the distributions should be compared to relevant data as a consistency check and, if time and funding allow, a subject matter expert should be consulted. Consistent with good engineering practice, the time to discover problems with data is before the analysis is complete.