Inference and robust extremes: Jonathan Jalbert
From Belle Taylor on June 28th, 2021
This talk has been automatically captioned. You can remove these by pressing CC on the video toolbar.
Name: Jonathan Jalbert
Talk Title: Extended generalized Pareto for subasymptotic tail analysis with an application to heatwave intensities
Abstract: For environmental applications, tail distribution plays a major role as it is often associated with extreme risks. A classical approach to characterize the tails is to model the exceedances above a sufficiently high threshold with the generalized Pareto distribution. For some applications, the choice of the threshold is difficult and the asymptotic conditions may not be satisfied. Extended generalized Pareto models can then be used when the threshold is not high enough. However, the existing extensions have an infinite or zero density at the origin, which makes them unsuitable for modeling the exceedances. A new extension of the extended generalized Pareto distribution is developed to model the exceedances above a sub-asymptotic threshold having a finite density and not zero at the threshold. The proposed extension provides a better estimate of extreme quantiles than existing models, especially for small samples. Finally, this new distribution is used to model heatwave intensities.
This talk is a contributed talk at EVA 2021.