Two Parts makes up this thesis. The first part is considering the asymptotic expansion and the choice of optimal sample fraction of a kind of Pickands type estimators with negative extreme value index under second regularly varying conditions.In the second part of this paper, a new kind of smoothing estimator for the positive extreme value index is proposed, i.e.whereand t,s∈R,t> s≥1,[x] denotes the smallest integer greater than or equal to x. The asymptotic properties of av(?)k,n, are obtained by using weak convergence of tail empirical process. Finite sample simulation shows that the proposed smoothing positive index estimator has fine properties compared to non-smoothing estimator (?)k,n and other smoothing estimators.
|