Seminar overview

We present the design and analysis of a randomized non-inferiority trial in talaromycosis, a major cause of human immunodeficiency virus (HIV)–related death in South and Southeast Asia. The talk will present the entire history of the trial highlighting the importance of a close collaboration between clinicians and statisticians. Moreover, it will focus on several statistical topics which arose during the trial including the choice of the non-inferiority margin, the issue of non-proportional hazards, and joint modeling of longitudinal fungal counts and mortality.

Graphical models based on Directed Acyclic Graphs (DAGs) represent a powerful tool for investigating dependencies among variables. It is well known that one cannot distinguish between DAGs encoding the same set of conditional independencies (Markov equivalent DAGs) using only observational data. However, the space of all DAGs can be partitioned into Markov equivalence classes, each being represented by a unique Essential Graph (EG), also called Completed Partially Directed Graph (CPDAG). In some fields, in particular genomics, one can have both observational and interventional data, the latter being produced after an exogenous perturbation of some variables in the system, or from randomized intervention experiments. Interventions destroy the original causal structure, and modify the Markov property of theunderlying DAG, leading to a finer partition of DAGs into equivalence classes, each one being represented by an Interventional Essential Graph (I-EG) (Hauser and Buehlmann). In this talk we consider Bayesian model selection of EGs under the assumption that the variables are jointly Gaussian. In particular, we adopt an objective Bayes approach, based on the notion of fractional Bayes factor, and obtain a closed form expression for the marginal likelihood of an EG. Next we construct a Markov chain to explore the EG space under a sparsity constraint, and propose an MCMC algorithm to approximate the posterior distribution over the space of EGs. Our methodology, which we name Objective Bayes Essential graph Search (OBES), allows to evaluate the inferential uncertainty associated to any features of interest, for instance the posterior probability of edge inclusion. An extension of OBES to deal simultaneously with observational and interventional data is also presented: this involves suitable modifications of the likelihood and prior, as well as of the MCMC algorithm. We conclude by presenting results for simulated and real experiments (protein-signaling data).
This is joint work with Federico Castelletti, Stefano Peluso and Marco Della Vedova (Universita' Cattolica del Sacro Cuore).

The histogram is widely used as a simple, exploratory display of data, but it is usually not clear how to choose the number and size of bins for this purpose. We construct a confidence set of distribution functions that optimally address the two main tasks of the histogram: estimating probabilities and detecting features such as increases and (anti)modes in the distribution. We define the essential histogram as the histogram in the confidence set with the fewest bins. Thus the essential histogram is the simplest visualization of the data that optimally achieves the main tasks of the histogram. We provide a fast algorithm for computing a slightly relaxed version of the essential histogram, which still possesses most of its beneficial theoretical properties, and we illustrate our methodology with examples. This is a joint work with Axel Munk, Hannes Sieling, and Guenter Walther.

We propose a model to forecast very large realized covariance matrices of returns, applying it to the constituents of the S&P 500 on a daily basis. To address the curse of dimensionality, we decompose the return covariance matrix using standard firm-level factors (e.g., size, value and profitability) and use sectoral restrictions in the residual covariance matrix. This restricted model is then estimated using vector heterogeneous autoregressive (VHAR) models estimated with the least absolute shrinkage and selection operator (LASSO). Our methodology improves forecasting precision relative to standard benchmarks and leads to better estimates of the minimum variance portfolios.

Accurate and reliable predictions of infectious disease dynamics canbe valuable to public health organizations that plan inter-ventions to decrease or prevent disease transmission. Generally seen as the most robust type of predictive models, ensemble-based methodologies combine outputs from individual models to create a combined prediction for a target of interest. We have implemented ensemble methods that form a predictive density for a target of interest as a weighted sum of the predictive densities from component models. In the simplest case,equal weight is assigned to each component model; in more complex cases, the weights can vary with the location, prediction target, week of the season when the predictions are made, a measure of component model uncertainty, recent observations of disease incidence, and other observed covariates. In this talk, I will describe the methods used to estimate thecomponent model weights for these weighted density ensembles. For simple settings we use the degenerate EM algorithm, and in more complex settings we use gradient tree boosting to estimate penalized weights as functions of covariates. Additionally, I will describe our evaluation of these methods in two applications of forecasting influenza in the US. In one application, we combined 21 models from 4 different research groups to create real-time ensemble forecasts of influenza in the US in the 2017/2018 winter flu season.

We consider the task of predicting whether loans are paid back or not. An often encountered problem in default prediction is the fact that there is relatively little default data since bankruptcies are usually uncommon events. We show how this issue can be alleviated by using a tree-boosted Tobit model in cases where there is additional data for the non-default events that is related to the default mechanism. Such additional data can consist of, for instance, number of days of delay by which loans were paid back, stock returns, or distance to default measures. We apply our proposed model for predicting defaults on loans made to Swiss small and medium-sized enterprises and obtain a large improvement in predictive performance compared to other state-of-the-art approaches.

Recognition of persons by their fingerprints is rather ubiquitous: be it for unlocking smartphones, border control and for forensic applications, or for access to medical services, as for example in India. Challenges are manifold: handling bad quality, partial observations, tampered prints, distortions due to imprinting, as well as growth of children and juveniles. In this talk we discuss fingerprint features and statistical methods for their extraction, modeling, simulation, quality estimation and growth. Their foundations range from image analysis over shape analysis to complex analysis.

tba