[Turing-Southampton] S3RI seminars this week

[Helen Ogden]

Dear all,

We have a packed schedule of S3RI seminars this week, with S3RI short 
talks on Thursday, and a seminar from Chris Skinner (LSE) on Friday. 
There is no S3RI seminar next week, so you will have time to recover!

On Thursday (21 February), from 2-3:30pm in 54 / 7035 (7B), we have 
short talks from some of our own:

14:00: Joanne Ellison on "Forecasting of cohort fertility under a 
hierarchical Bayesian approach"
14:20: Samuel Jackson on "Approaches to the emulation of chains of 
computer models with application to epidemic policy making"
14:40: Refreshments in the staff reading room on level 4 of building 54
15:00: Andrew Hinde on "Mortality statistics by place and cause of death 
in England and Wales, 1851-1910"

The short talks will also be available via a live web-cast at
https://coursecast.soton.ac.uk/Panopto/Pages/Viewer.aspx?id=143d70f7-eba8-4fd1-a191-a9f8008caf43

On Friday (22 February) at 2pm in 54 / 8033 (8B), we have an S3RI 
seminar from Chris Skinner (LSE) on "Informative cluster size and 
multilevel models".

Details of all talks are given below.

All are welcome!

Best wishes,

Helen

**Thursday 21 Feb, 2-3:30pm, 54/7035**

Forecasting of cohort fertility under a hierarchical Bayesian approach

Joanne Ellison, University of Southampton

In keeping with the recent literature, we propose a hierarchical 
Bayesian model to forecast cohort fertility rates. Using Hamiltonian 
Monte Carlo methods and a dataset from the Human Fertility Database, we 
obtain forecasts for 30 countries and use scoring rules to 
quantitatively assess their predictive accuracy.


Approaches to the emulation of chains of computer models with 
application to epidemic policy making

Samuel Jackson, University of Southampton

We describe novel Bayes linear emulation methodology to analyse chains 
of computer models, where the outputs of one model feed into the next 
model. We focus on emulating each model in the chain individually, thus 
we have developed emulators for models where the input is uncertain (as 
the inputs to all but the first model are the uncertain emulated outputs 
of another model). The first method proposes analysing each emulator’s 
behaviour for a sample of inputs arising from a probabilistic 
distribution commensurate with our beliefs about the output of the 
previous emulator. The second method extends the field of emulation to 
directly incorporate uncertain inputs within each emulator itself. We 
demonstrate the potential of these novel emulation approaches using 
intuitive examples, before demonstrating their advantage over the single 
emulator approach for modelling of epidemic diseases.


Mortality statistics by place and cause of death in England and Wales, 
1851-1910

Andrew Hinde, University of Southampton

This presentation will describe the challenges in producing a set of 
regional mortality statistics by place and cause of death for England 
and Wales between 1851 and 1910. The challenges arise from the fact that 
(1) during the period the Registrar General made several changes to the 
classification scheme used to report causes of death, and (2) population 
growth and change meant that the geographical units used in the 
reporting varied over time. The paper is an historical illustration of 
issues faced by many statisticians who need to work with long time 
series of official data.


**Friday 22 Feb, 2-3pm, 54/8033**

Informative cluster size and multilevel models

Chris Skinner, LSE

When using multilevel modelling to analyse clustered data, there may be 
reasons why the sizes of clusters are correlated with cluster-level 
random effects. If so, cluster size is called informative and 
conventional estimation of the multilevel model parameters may be 
biased. This talk will introduce the idea of informative cluster size, 
mentioning its relevance to marginal models as well as multilevel 
models. A novel within-cluster resampling approach to bias correction 
will be described.


For the current schedule of S3RI seminars, see 
https://tinyurl.com/s3riseminar



-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mailman.ecs.soton.ac.uk/pipermail/turing-southampton/attachments/20190218/d319ff31/attachment.html