Recherche

The research areas in which the members of the Institute of Statistics, Biostatistics and Actuarial Sciences are working are diverse.  The main areas of expertise are non- and semi-parametric regression techniques, time series analysis, survival analysis, medical and industrial statistics, extreme value analysis and statistics for the actuarial sciences, as well as insurance and financial mathematics.

In the context of non- and semi-parametric regression, the Institute is a leading expert in the area of frontier estimation and particularly in the application of non- and semi-parametric approaches for this problem in the context of efficiency analysis. A lot of research is also carried out in the context of inverse problems in econometrics, and its applications in instrumental regression and deconvolution problems. The study of semi-parametric regression models (e.g. single index models, partial linear models) and of non-parametric location-scale models, is another area in which the Institute is taking a leading role.  More generally, a lot of activity can be summarized by the development of up-to-date methodology for denoising statistical signals in one and in higher dimensions. Modern non-linear methods, among others based on wavelets, do not only serve to this end but also for the development of functional data and image analyses and clustering.

The analysis of time series is a second cornerstone of the research activities at the Institute.  The focus lies on the modelling and analysis of non-stationary time series, multivariate (high-dimensional) time series, factor models, volatility models, spectral density estimation and goodness-of-fit methods.  Moreover, applications in statistical signal processing, and biomedical, economic and financial time series are studied. 

The analysis of data coming from medical or industrial studies is a further research topic to which much attention is paid at the Institute.  Medical data are often subject to censoring (survival analysis).  The non- and semi-parametric modelling of this type of data is studied in detail, both the asymptotics for these models, as the application to medical data. Moreover, the study of clinical trials gives rise to developping models, both frequentist and bayesian, for biological processes, but this applies also to the context of industrial statistics, e.g. in chemometrics, and in quality control. More specifically, the focus lies on experimental design and multicriteria optimisation with applications in drug discovery, and on the analysis and modelling of time intensity curves in sensometrics.

The quantitative analysis of financial and insurance risks is intended to help economic agents to design efficient strategies for managing these risks. The ISBA research team has developed a widely recognized expertise in that field, especially in stochastic orderings and inequalities, financial econometrics, dependence modeling, in particular by means of copulas, actuarial risk theory, mathematical finance, and extreme value theory: modelling of extremes in univariate and multivariate time series, and in particular in Markov chains.

 

 

.