B1673
Title: The role of resampling methods in extreme value parameters estimation
Authors: Dora Prata Gomes - NOVA.ID.FCT FCT-UNL (Portugal) [presenting]
Manuela Neves - FCiencias.ID, Associacao para a Investigacao e Desenvolvimento de Ciencias (Portugal)
Abstract: Extreme value theory (EVT) has many applications in different areas such as flooding, rainfall, and insurance claims, for example. Several researchers have applied EVT to obtain more reliable estimates of extreme events. Climate change has brought in unprecedented way new weather patterns, one of which is changes in extreme rainfall. In this example, to build a resilient society and achieve sustainable development, it is paramount that adequate inference about extreme rainfall be made. EVT provides analogues of the central limit theorem for the extreme values in a sample. According to the central limit theorem, the mean of a large number of random variables, irrespective of the distribution of each variable, is distributed approximately according to a Gaussian distribution. For example, the sea surface elevation is often modelled as a sum of several individual random waves and accordingly its distribution is often assumed to be Gaussian. According to extreme value theory, the extreme values in a large sample have an approximate distribution that is independent of the distribution of each variable. Some challenges have been developed by EVT to obtain more reliable extreme value parameter estimates. Resampling procedures such as the bootstrap have been used to improve parameter estimation in EVT. New approaches, based on bootstrap procedures are shown and are illustrated with a real data set using the R software.
