| 2007 | |
|---|---|
|
5.00 p.m., 9 May 2007 |
J.O. Ramsay, G. Hooker, D. Campbell and J. Cao (McGill University) Parameter Estimation for Differential Equations: A Generalized Smoothing Approach Dynamic systems models involve differential equations that are seldom analytically solvable, so that the tools of statistics have been difficult to apply to data-fitting problems. A generalized smoothing approach involving a multilevel parameter structure and using a profiling strategy permits parameter and confidence region estimation in a wide range of dynamic model and data configurations.
Electronic version of the paper: [PDF]. |
| 2007 | |
|---|---|
|
5.00 p.m., 18 April 2007 |
P. Diggle (Lancaster University) D. Farewell (Cardiff University) R. Henderson (University of Newcastle upon Tyne) Analysis of Longitudinal Data with Drop-Out: Objectives, Assumptions and a Proposal When analysing longitudinal measurements terminated by drop-out, we argue that existing inferential objectives can be too vague, existing missing-data assumptions unrealistic and existing models overly complicated. We review standard approaches in the light of three specific objectives, before proposing an easily implemented new approach based on a dynamic incremental model.
Electronic version of the paper: [PDF]. |
|
5.00 p.m., 31 Jan 2007 |
D. ZENG (University of North Carolina) D. Y. LIN (University of North Carolina) Semiparametric regression models with censored data
New classes of semiparametric models are proposed for the regression
analysis of censored data. These include heteroscedastic transformation
models for survival data, random-effects transformation models for
multivariate failure time data and joint transformation models for
repeated measures and event times. Inference procedures and numerical
algorithms based on nonparametric likelihood are described. Applications
to medical studies are provided. |
| 2006 | |
|---|---|
|
5.00 p.m., 11 Oct 2006 |
MARK S. HANDCOCK (University of Washington) ADRIAN E. RAFTERY (University of Washington) JEREMY M. TANTRUM (University of Washington) Model based clustering for social networks
Network models are widely used to represent relations among interacting units or actors.
Network data often exhibit transitivity, meaning that two actors that have ties to a third
actor are more likely to be tied than actors that do not, homophily by attributes of the
actors or dyads, and clustering. Interest often focuses on finding clusters of actors or ties,
and the number of groups in the data is typically unknown. We propose a new model, the
Latent Position Cluster Model (LPCM), under which the probability of a tie between two
actors depends on the distance between them in an unobserved Euclidean social space,
and the actors' locations in the latent social space arise from a mixture of distributions, each
one corresponding to a cluster. We propose two estimation methods: a two-stage maximum
likelihood method, and a Bayesian MCMC method; the former is quicker and simpler, but the
latter performs better. We also propose a Bayesian way of determining the number of clusters
present using approximate conditional Bayes factors. It models transitivity, homophily by
attributes and clustering simultaneously, and does not require the number of clusters to be
known. The model makes it easy to simulate realistic networks with clustering, potentially
useful as inputs to models of more complex systems of which the network is part, such as
epidemic models of infectious disease. We apply the model to two networks of social relations.
|