|
|
Royal Statistical Society
|
| 2001 | |
|---|---|
|
5.00 p.m., 12 December |
STEFFEN LAURITZEN (Aalborg University) and
THOMAS RICHARDSON (University of Washington)
Chain graph models and their causal interpretations Chain graphs are a natural generalization of directed acyclic graphs (DAGs) and undirected graphs. However, the apparent simplicity of chain graphs belies the subtlety of the conditional independence hypotheses that they represent. There are a number of simple and apparently plausible, but ultimately fallacious, interpretations of chain graphs that are often invoked, implicitly or explicitly. These interpretations also lead to flawed methods for applying background knowledge to model selection. We present a valid interpretation by showing how the distribution corresponding to a chain graph may be generated from the equilibrium distributions of dynamic models with feedback. These dynamic interpretations lead to a simple theory of intervention, extending the theory developed for DAGs. Finally, we contrast chain graph models under this interpretation with simultaneous equation models which have traditionally been used to model feedback in econometrics. Electronic version of the paper (and erratum). |
| 2002 | |
|
5.00 p.m., 13 February |
YINGCUN XIA (University of Hong Kong and University of Cambridge),
HOWELL TONG (University of Hong Kong and LSE),
WK LI (University of Hong Kong) and
LI-XING ZHU (University of Hong Kong and Chinese Academy of Sciences)
An Adaptive Estimation of Dimension Reduction Space In regression, how to best approximate the regression function or view high dimensional data in a (hopefully much) lower dimensional subspace is of primary importance. We present an approach which makes no parametric assumption on the functional form of the regression surface, does not restrict the regressors to be independent or to follow any particular type of distribution, and requires no under-smoothing of the nonparametric estimate of the unknown regression function. Electronic version of the paper. Datasets and programs used in the the paper. |
|
5.00 p.m., 13 March |
DAVID SPIEGELHALTER (MRC Biostatistics Unit, Cambridge),
NICOLA BEST (Imperial College, London),
BRADLEY CARLIN (University of Minnesota) and
ANGELIKA VAN DER LINDE (University of Bremen)
Bayesian measures of model complexity and fit We investigate the estimation of the effective number of parameters, denoted pD, in a Bayesian model by the difference between the posterior mean of the deviance and the deviance at the posterior means of the parameters of interest. A Deviance Information Criterion (DIC) may then be obtained by adding the 'complexity' pD to the 'fit', taken as the posterior mean deviance. These quantities are easy to compute using MCMC methods and their use in comparing complex Bayesian models is illustrated. Electronic version of the paper. |
|
2.00 p.m., 22 May |
Extended Ordinary Meeting on STATISTICAL MODELLING AND ANALYSIS OF GENETIC DATA Further details are available here. |