نشریه اندیشه آماری
سال یازدهم شماره 2 (پیاپی 22، پاییز و زمستان 1385)

In this paper، we introduce the half-logistic distribution. The moments، skewness and kurtosis coe±cients re obtained. The maximum likelihood method does not provide an explicit estimator for the scale parameter. We provides a simple method of deriving an explicit estimator by approximating the likelihood function and study the e±ciency of this method by simulation.

The univariate and multivariate normal distributions have a central role in statistics. However there are many processes that do not follow from normal law, so there is a need for statistical modeling by the use of multivariate nonnormal distributions. In recent decades among nonnormal multivariate distributions the spherical and elliptical distributions more than other distributions have attracted much attention, because they have useful properties similar to multivariate normal distributions. In this paper we in- troduce elliptical distributions, it's properties and uniform test for goodness-of-¯t test of elliptical distributions.

In recent years, there have been considerable developments of models for time-dependent data. In this paper, with emphasis on counts, two classes of models which apply time-dependent data modeling have been de¯ned. These two classes were ¯rst suggested by Cox (1981) and include observation driven and parameter driven mod- els. These classes and some of their properties are introduced and parameter estimation methods are mentioned in short. Finally, in order to show their vast applications, we have applied one of the members of observation driven models as GLARMA to model the number of car collision in Mashhad.

In this paper we give a general solution to the problem of ¯nding UMVUE for function of the unknown end points of the support of a uniform distribution.

In Momeni and Alamtsaz (1384), in°ated-parameter negative binomial and poisson distributions were studied as generalizations of the corresponding classic distri- butions. These distributions are in fact two members of the class of in°ated-parameter power series distributions which have extensive applications in modeling correlated count data, particularly in Econometrics and Insurance and are studied by Minkova(2002). In this article, the remaining important distributions of the class, i.e., Bernoulli, binomial and log-series distributions are studied. Then, their pmf.'s, pgf.'s and their uni¯ed forms are presented. Finally, we shall illustrate their applications by a real data set in connec- tion with number of insurance claims.

In this paper seasonal adjustment and components of time series are con- sidered.It also provides with an example, a brief summary of the algorithm,tests and options of the X12-ARIMA method by SAS software.