B0346
Title: A study on extremal index robust estimation considering it like a proportion
Authors: Manuela Souto de Miranda - University of Aveiro (Portugal) [presenting]
M Cristina Miranda - University of Aveiro (Portugal)
Conceicao Amado - Universidade de Lisboa (Portugal)
Ivette Gomes - FCiencias.ID, Universidade de Lisboa and CEAUL (Portugal)
Abstract: Under specific conditions, the limit distribution of the maxima of stationary sequences can exist, even in the presence of some dependence structures. Then, the degree of dependence might be studied through a parameter of the Extreme Value Distribution family, named the extremal index (EI). That parameter can be interpreted in different contexts, namely, as the numerical inverse of the mean size of clusters of exceedances, or as the multiplicity of a compound Poisson point process. Generally, EI estimation methods are focused on the mean clusters size estimation in the limit distribution. The inverse of that estimate provides the extremal index estimate. We investigate a different approach based on robust estimation of the parameter by itself, interpreted as a proportion. The procedure takes into account the distribution of the inter-exceedances times of the process. A simulation study was carried out to perform a comparative analysis of both approaches.
