Dear all,
RSS Lancashire and East Cumbria local group will be holding a seminar at 4pm on Thu 1 Dec in A54, Postgraduate Statistics Centre at Lancaster University. The meeting will be preceded by a very brief AGM of the local group. It will be followed by a wine reception and refreshments.
Title: Integrating computer simulations, emulation and extreme value methods to quantify uncertainty in risk models
Speaker: Marc Kennedy, The Food and Environment Research Agency
Abstract:
Quantitative risk assessment usually involves estimating small probabilities of rare, high consequence events. Relative uncertainty is inevitably high due to extremely limited data. Computer simulations can provide additional information but realistic stochastic models require massive run times to generate useful information about the rare events of interest. Efficient methods are therefore required, particularly if the risk model is complex.
In many risk scenarios, the actual level of the output quantity is also important, to understand how serious the problem could really be. A typical example would be the proportion of individuals exposed to a chemical above a fixed threshold, but also by how much high risk individuals within that population might exceed the threshold, to compare with the 'safety factor' that is built into the threshold choice. Risk managers must ensure that a population is sufficiently protected by accounting for multiple sources of variability. The distribution tail then becomes important, but the standard approach of fitting lognormal distributions is inadequate. The combined use of computer model emulation and extreme value techniques will be discussed to deal with non-Gaussian stochastic outputs and to quantify uncertainty in the output distribution tail. These methods can also be used to quickly identify those model parameters having the greatest impact on the tail, and therefore the probabilities of rare events.
Example applications related to occupational and environmental exposure will be presented.
*******************
Dates for your diary, 2012:
Thu 23 Feb, 4pm:
Chris Glasbey, Biomathematics and Statistics Scotland
Spatio-temporal weather models
Venue: Lancaster University
Thu 22 Mar, 4pm:
Ashley Jones, Alder Hey Children's NHS Foundation Trust
The use of systematic reviews in the design of randomised controlled trials
Venue: University of Central Lancashire
You may leave the list at any time by sending the command
SIGNOFF allstat
to [log in to unmask], leaving the subject line blank.
|