Bayesian duality and risk analysis on the statistical manifold of exponential family with censored data (Q1639569): Difference between revisions

From MaRDI portal
Import240304020342 (talk | contribs)
Set profile property.
Set OpenAlex properties.
Property / full work available at URL
 
Property / full work available at URL: https://doi.org/10.1016/j.cam.2018.03.043 / rank
 
Normal rank
Property / OpenAlex ID
 
Property / OpenAlex ID: W2800009245 / rank
 
Normal rank

Revision as of 02:25, 20 March 2024

scientific article
Language Label Description Also known as
English
Bayesian duality and risk analysis on the statistical manifold of exponential family with censored data
scientific article

    Statements

    Bayesian duality and risk analysis on the statistical manifold of exponential family with censored data (English)
    0 references
    0 references
    0 references
    0 references
    13 June 2018
    0 references
    The authors investigate Bayesian duality on the statistical manifold of the exponential family with life testing data. The experimental planning and statistical manifold of the exponential family of distributions are discussed after which the generalized progressive type-I censoring scheme with continuous monitoring is described. Based on the observed experimental data, the statistical manifold is constructed with a new cumulant generating function and it is shown that the manifold has a new free energy convex function. Based on the constructed manifold, the Bregman divergence between two parameter points is studied and the dual coordinate system, dual function and the dualistic structure on the manifold are obtained. The MLE can then be obtained by minimizing the Bregman divergence induced from the dual function. The Bayesian duality on the statistical manifold is investigated and the Bayesian estimators of the parameter and its dual are provided. Prior selection with both the conjugate informative prior and Jeffreys and reference non-informative priors is discussed and the Bayesian prediction of the unobserved data is discussed and the prediction is evaluated using risk function based on different divergences as loss function. The Gamma distribution is chosen as an example to illustrate the process to obtain the dual coordinate system and the dual function. The corresponding estimators are also derived. A real data set on a prostate cancer survival study is analyzed in support of the methodologies and the experimental schemes.
    0 references
    Bayesian duality
    0 references
    statistical manifold
    0 references
    risk analysis
    0 references
    Bregman divergence
    0 references
    censored data
    0 references

    Identifiers

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