Dans Beirlant et Guillou [1] un modèle de régression exponentiel basé sur l'écart du logarithme de statistiques d'ordres consécutives d'un échantillon issu d'une loi de type Pareto a été introduit en présence de censure. De cette représentation, ils obtiennent un estimateur de l'index de Pareto. Dans cette note, nous revisitons cette adaptation de l'estimateur de Hill [5] en établissant en particulier sa convergence presque sûre sous des conditions très générales sur le nombre Nr de données non censurées.
In Beirlant and Guillou [1] an exponential regression model was introduced on the basis of scaled log-spacing between subsequent extreme order statistics from a Pareto-type distribution in the presence of censoring. From this representation, they derived an estimator for the Pareto index. In this note, we revisit this adaptation of the popular Hill [5] estimator for heavy-tailed distributions, generalizing the almost sure convergence of this estimator under very general conditions on Nr, the number of non-censored observations.
Révisé le :
Publié le :
@article{CRMATH_2002__335_4_375_0, author = {Delafosse, Emmanuel and Guillou, Armelle}, title = {Almost sure convergence of a tail index estimator in the presence of censoring}, journal = {Comptes Rendus. Math\'ematique}, pages = {375--380}, publisher = {Elsevier}, volume = {335}, number = {4}, year = {2002}, doi = {10.1016/S1631-073X(02)02486-X}, language = {en}, url = {http://archive.numdam.org/articles/10.1016/S1631-073X(02)02486-X/} }
TY - JOUR AU - Delafosse, Emmanuel AU - Guillou, Armelle TI - Almost sure convergence of a tail index estimator in the presence of censoring JO - Comptes Rendus. Mathématique PY - 2002 SP - 375 EP - 380 VL - 335 IS - 4 PB - Elsevier UR - http://archive.numdam.org/articles/10.1016/S1631-073X(02)02486-X/ DO - 10.1016/S1631-073X(02)02486-X LA - en ID - CRMATH_2002__335_4_375_0 ER -
%0 Journal Article %A Delafosse, Emmanuel %A Guillou, Armelle %T Almost sure convergence of a tail index estimator in the presence of censoring %J Comptes Rendus. Mathématique %D 2002 %P 375-380 %V 335 %N 4 %I Elsevier %U http://archive.numdam.org/articles/10.1016/S1631-073X(02)02486-X/ %R 10.1016/S1631-073X(02)02486-X %G en %F CRMATH_2002__335_4_375_0
Delafosse, Emmanuel; Guillou, Armelle. Almost sure convergence of a tail index estimator in the presence of censoring. Comptes Rendus. Mathématique, Tome 335 (2002) no. 4, pp. 375-380. doi : 10.1016/S1631-073X(02)02486-X. http://archive.numdam.org/articles/10.1016/S1631-073X(02)02486-X/
[1] Pareto index estimation under moderate right censoring, Scand. Actuar. J., Volume 2 (2001), pp. 111-125
[2] The law of the iterated logarithm for normalized empirical distribution function, Z. Wahrscheinlichkeitstheorie und Verwandte Gebiete, Volume 38 (1977), pp. 147-167
[3] Central limit theorems for sums of extreme values, Math. Proc. Cambridge Philos. Soc., Volume 98 (1985), pp. 547-558
[4] Almost sure convergence of the Hill estimator, Math. Proc. Cambridge Philos. Soc., Volume 104 (1988), pp. 371-381
[5] A simple general approach to inference about the tail of a distribution, Ann. Statist., Volume 3 (1975), pp. 1163-1174
[6] G. Matthys, E. Delafosse, A. Guillou, J. Beirlant, Estimating high quantiles, Technical Report, 2002
[7] Empirical Processes with Applications to Statistics, Wiley, New York, 1986
[8] Estimation of parameters and large quantiles based on the k largest observations, J. Amer. Statist. Assoc., Volume 73 (1978), pp. 812-815
Cité par Sources :