Estimating the Prediction Mean Squared Error in Gaussian Stochastic Processes with Exponential Correlation Structure (Q4956056)

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 1452469
Language Label Description Also known as
default for all languages
No label defined
    English
    Estimating the Prediction Mean Squared Error in Gaussian Stochastic Processes with Exponential Correlation Structure
    scientific article; zbMATH DE number 1452469

      Statements

      Estimating the Prediction Mean Squared Error in Gaussian Stochastic Processes with Exponential Correlation Structure (English)
      0 references
      0 references
      24 May 2000
      0 references
      computer experiments
      0 references
      exponential covariance
      0 references
      Gaussian stochastic process
      0 references
      information matrix
      0 references
      mean squared error
      0 references
      spatial linear model
      0 references
      spatial statistics
      0 references
      The problem of estimating the squared prediction error when predicting the process at unobserved locations is considered. Namely, a real-valued Gaussian stochastic process \(\{W(t); t\in [0,1]^{d}\}\) satisfying the model \[ W(t)=f(t)^{\text{T}}\beta +Z(t)+\varepsilon(t) \] is considered, where \(f(t)\) is a known vector regression function and \(\beta\) is a vector of unknown parameters. The stochastic processes \(Z(t)\) and \(\varepsilon(t)\) are assumed to be independent with zero mean and variance \(\sigma^{2}>0\) and \(\tau^{2}>0,\) respectively. The process \(\varepsilon(t)\) reflects uncorrelated measurement error whereas \(Z(t)\) denotes the systematic deviation of \(W(t)\) from the mean \(f(t)^{\text{T}}\beta.\) An approximation taking into account the additional variability due to estimating parameters involved in the correlation structure was developed by \textit{R. N. Kackar} and \textit{D. A. Harville} [J. Am. Stat. Assoc. 79, 853-862 (1984; Zbl 0557.62066)] and was revisited by \textit{D. A. Harville} and \textit{D. R. Jeske} [ibid. 87, No. 419, 724-731 (1992; Zbl 0763.62039)] as well as by \textit{D. L. Zimmerman} and \textit{N. Cressie} [Ann. Inst. Stat. Math. 44, No. 1, 27-43 (1992; Zbl 0760.62090)]. The present paper discusses an extension of these methods. The approaches are compared via an extensive simulation study for models with and without random error term. The results show that considering the additional variability in the predictor due to estimating the covariance structure is of great importance and should not be neglected in practical applications.
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references