Prof. Jonathan Wakefield

Jonathan is Professor of Statistics and Biostatistics at the University of Washington. His research interests include the analysis of complex survey data, small area estimation, spatial epidemiology, space-time models for infectious disease data, software development, estimation of excess mortality, estimating national and subnational mortality burden including child mortality, ecological inference for non-infectious and infectious disease data and the links between Bayes and frequentist procedures. His is a Fellow of the American Statistical Association and was awarded the Guy Medal in Bronze by the Royal Statistical Society in 2000.

Small Area Estimation in Low- and Middle-Income Countries

Small Area Estimation (SAE) in Low- and Middle-Income Countries (LMICs) is fundamentally more difficult than in high income countries, in part because of the lack of good auxiliary information, with census data often being unreliable and/or outdated. In this talk I will discuss models and software for carrying out SAE in LMICs, using examples from the work of my collaborators and I with the United Nations (UN) and the World Health Organization (WHO). These examples include child mortality, vaccination coverage, female educational attainment and fertility. I will discuss the pros and cons of both area-level and unit-level models. Area-level (Fay-Herriot) models rely on less assumptions but require sufficient data to obtain a weighted estimator and an associated variance. Unit-level models need to acknowledge the survey design, and require a tricky aggregation step but can be used with sparse data. I will discuss the nature of spatial smoothing, and strategies for the avoidance of over-smoothing. Other topics to be discussed will include benchmarking, variance modeling, combining different data sources, and the use of machine learning models within Bayesian SAE models.

Details

The talk will take place on Thursday 29th May at 11am in MS.02. Lunch will be provided afterwards in the The Street (the Zeeman Atrium).

Prof. Nicolas Chopin

Nicolas is Professor of Data Sciences/Statistics/Machine Learning at ENSAE, Institut Polytechnique de Paris. He is a world-renowned expert on Bayesian computation, and author of the book "An introduction to Sequential Monte Carlo", with Omiros Papaspiliopoulos.

Saddlepoint Monte Carlo and its Application to Exact Ecological Inference

Assuming X is a random vector and A a non-invertible matrix, one sometimes need to perform inference while only having access to samples of Y = AX. The corresponding likelihood is typically intractable. One may still be able to perform exact Bayesian inference using a pseudo-marginal sampler, but this requires an unbiased estimator of the intractable likelihood.

We propose saddlepoint Monte Carlo, a method for obtaining an unbiased estimate of the density of Y with very low variance, for any model belonging to an exponential family. Our method relies on importance sampling of the characteristic function, with insights brought by the standard saddlepoint approximation scheme with exponential tilting. We show that saddlepoint Monte Carlo makes it possible to perform exact inference on particularly challenging problems and datasets. We focus on the ecological inference problem, where one observes only aggregates at a fine level. We present in particular a study of the carryover of votes between the two rounds of various French elections, using the finest available data (number of votes for each candidate in about 60,000 polling stations over most of the French territory).

We show that existing, popular approximate methods for ecological inference can lead to substantial bias, which saddlepoint Monte Carlo is immune from. We also present original results for the 2024 legislative elections on political centre-to-left and left-to-centre conversion rates when the far-right is present in the second round. Finally, we discuss other exciting applications for saddlepoint Monte Carlo, such as dealing with aggregate data in privacy or inverse problems.

joint work with: Théo Voldoire (Harvard), Guillaume Rateau (INSEE) and Robin J. Ryder (Imperial)

Details

The talk will take place on Tuesday 18th March at 11am in MS.02. Lunch will be provided afterwards in the The Street (the Zeeman Atrium).

Prof. Sofia Olhede

Sofia is Chair of Statistical Data Science at EPFL. She is a Fellow of the IMS for "seminal contributions to the theory and application of large and heterogeneous networks, random fields and point process, for advancing research in data science”.

Modelling multiplex networks

Multiple networks are complex and highly non-Euclidean objects. This talk will focus on what is known as multiplex networks, representing multiple interaction between the same set of objects, but where the interactions may be of several different types. We shall discuss how a graph limit framework can be used for such objects, and how to naturally characterise the complexity of the whole system of interactions. This characterisation will be used on complexity measures introduced in information theory, and the notion of entropy. This gives us a tool to generally think about systems of relationship.

Details

The talk will take place on Monday 9th December at 11am in MS.03. Lunch will be provided afterwards in the The Street (the Zeeman Atrium).