利用顺序统计量分组,本论文提出了一类新的重尾分布的极值指数估计量,并在适当的正规变换条件下讨论了该估计量的强弱相合性及渐近正态性。
This paper proposes a new kind of index estimator of a heavy-tailed distribution when only a few largest values are observed within blocks. The asymptotic properties, such as the weak and strong consistency and the asymptotic normality of this kind of tail index estimator have also been considered under suitable regularly varying conditions.