Showing papers in "Scandinavian Journal of Statistics in 1985"

A class of distributions which includes the normal ones

Abstract: On introduit une nouvelle classe de fonctions de densite dependant du parametre de forme λ, telles que λ=0 corresponde a la densite normale standard

The moment structure of ARCH processes

Abstract: On considere les aspects de serie temporelle d'une classe de modeles introduits par Engle (1982)

On reliability theory and its applications

Abstract: On analyse plusieurs resultats de la theorie de la fiabilite ainsi que leur utilite dans les applications industrielles

Kernel estimators of zeros and of location and size of extrema of regression functions

Abstract: On considere des estimateurs du noyau de fonctions de regression dans le modele de regression a plan fixe

A theorem for determining the compensator of a counting process

Abstract: Let {N(t)} be a counting process adapted to a history {Y,} and assume that there exists a process {i(t)} such that 1(t)=lim E(N(t+h.)-N(t) I |)/h,. hn 40 Under certain conditions we prove that {N(t)-fh )A(s) ds} is an X, martingale, i.e. {f'o(s) ds} is the compensator (often called the dual predictable projection) of {N(t)}. Our result is closely related to Dolivo's (1974) Theorem 2.5.1. The proof is very much like Dolivo's proof of the (b) part of this theorem. Some mistakes are pointed out in Dolivo's theorem/proof.

On robust kernel estimation of derivatives of regression functions

Abstract: On propose une methode derivee de l'estimation M (Huber, 1981), et on constate que l'estimation du noyau robuste de la premiere derivee est un noyau ordinaire operant sur des residus transformes convenables

Saddlepoint Formulas for Reproductive Exponential Models

Abstract: The renormalized saddlepoint approximation for the probability density function of the empirical mean x of a sample xl, . . . , x, from a particular type of reproductive exponential model is shown to be exact for all n=1, 2, .... This result implies that the only reproductive exponential models of this kind for one-dimensional variates are the models corresponding to the normal, the inverse Gaussian and the gamma distributions.

The null distribution for a test of constant versus «bathtub» failure rate

Abstract: On obtient la distribution exacte de la statistique test de Bergman (1979) sous l'hypothese d'exponentialite

Distribution-free procedures for location and symmetry inference problems based on the empirical characteristic function

Abstract: On propose des methodes non parametriques, basees sur la fonction caracteristique empirique, pour estimer le centre δ d'une distribution symetrique et pour tester l'hypothese de symetrie autour d'une mediane inconnue

Uniform Confidence Bounds for Regression Based on a Simple Moving Average

Abstract: Let m(x)=E(YIX=x) be the regression function of Y on x. Suppose that YI, . . , Y. are independent observations of Y at x=xl,. . . x,x. We consider nearest neighbour estimates, *h(x), and employ well-known inequalities to obtain finite sample size uniform confidence bounds for Em(x) and asymptotic uniform confidence bounds for Em(x) and m(x) based on m(x). Finally we discuss bias and consistency properties of m(x).

Confidence limits from c|j|1/2 L in the single-parameter case

Abstract: On obtient des limites de confiance uni- et bilaterales, correctes jusqu'a l'ordre 0(n −3/2 ), conditionnelles et non conditionnelles

On some new ageing properties

Abstract: On introduit deux nouvelles classes de distributions de vie, SIFR et SNBU, ainsi que leurs duales

Delays reducing the waiting time in queueing systems with cyclic services

Abstract: Etude de strategies de commande pour des modeles decrivant le comportement de systemes de transport ou de reseaux d'ordinateur

A strong non-linear renewal theorem with applications to sequential analysis

Abstract: On etablit une approximation forte pour les temps de premier passage d'un processus stochastique a travers une barriere non lineaire. On donne des applications de ce resultat en analyse sequentielle

On testing hypotheses in a mixed linear model for incomplete block designs

Abstract: On cherche a demontRer que les estimateurs de l'analyse interblocs ou intrablocs possedent certaines proprietes desirees dans le modele lineaire mixte pour les plans de blocs. On donne un test exact pour des hypotheses concernant les contrastes de traitement

Asymptotic results for estimators in restricted randomization designs

Abstract: On donne un exemple de l'utilite de la theorie centrale limite des martingales pour les statistiques inferentielles