On Sequential Maximum Likelihood Estimation for Exponential Families of Stochastic Processes (Q3740849)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 3976140
Language Label Description Also known as
default for all languages
No label defined
    English
    On Sequential Maximum Likelihood Estimation for Exponential Families of Stochastic Processes
    scientific article; zbMATH DE number 3976140

      Statements

      On Sequential Maximum Likelihood Estimation for Exponential Families of Stochastic Processes (English)
      0 references
      0 references
      1986
      0 references
      minimal sufficient statistic
      0 references
      consistency
      0 references
      asymptotic normality
      0 references
      Bartlett adjustments
      0 references
      diffusion type process
      0 references
      likelihood function
      0 references
      discrete and continuous time parameter processes
      0 references
      noncurved exponential family
      0 references
      stopping time
      0 references
      Laplace transform
      0 references
      maximum likelihood estimator
      0 references
      likelihood ratio test statistic
      0 references
      counting processes
      0 references
      birth-death process
      0 references
      Statistical models for stochastic processes for which the likelihood function corresponding to permanent observation of the process between time 0 and a fixed time t has the form \[ L_ t(\theta)=a_ t \exp (\theta A_ t-f(\theta)B_ t),\quad \theta \in \Theta \subseteq R, \] where \(a_ t\), \(A_ t\), \(B_ t\) depend on the observed process, and f is a real deterministic function, are considered. Both discrete and continuous time parameter processes are studied. Sequential procedures for which the model is, exactly or approximately, a noncurved exponential family are investigated. The procedure is to observe the process between time 0 and the stopping time \(\tau (u)=\inf \{t:\alpha A_ t+\beta B_ t\geq u\}\), where u, \(\alpha\), \(\beta\) are known constants. The Laplace transform of \(A_{\tau}\) or \(B_{\tau}\) is given. Based on these expressions, asymptotic and exact results are derived about the distribution of \(A_{\tau}\) and \(B_{\tau}\), of the maximum likelihood estimator and of the likelihood ratio test statistic for a point hypothesis. The general results are applied to processes of diffusion type and counting processes. Extensions of the results to the classes of exponential type stochastic processes with n-dimensional parameters are also discussed. The results are applied to the birth-death process. In the paper a review of related results is given as well.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references