A Full Bayesian Non‐parametric Analysis Involving a Neutral to the Right Process

From MaRDI portal





Neutral to the right (NTR) random distribution functions \(F(x)\) on \((0,\infty)\) are considered as priors for nonparametric Bayesian analysis. NTR means that \(F(t)=1-\exp(-Z(t))\), where \(Z(t)\) is an independent increments process, non-decreasing and right-continuous a.s., \(Z(0)=0\), \(Z(t)\to +\infty\) as \(t\to+\infty\). Special attention is paid to the process \(Z\), which has the Lévy measure \[ dN_t(z)=dz(1-\exp(-z))^{-1}\int_0^t\exp(-Z\beta(s))d\alpha(s) \] (beta-Stacy process). Here \(\beta\) is a non-negative function and \(\alpha\) is a measure. If \(\alpha\) is a finite measure and \(\beta(t)=\alpha(t,\infty)\) the beta-Stacy process is a Dirichlet process. If the prior process is the Dirichlet one, then the posterior process is beta-Stacy if the data include right-censored observations. The problem of NTR process simulation is considered. An illustrative analysis is presented.











This page was built for publication: A Full Bayesian Non‐parametric Analysis Involving a Neutral to the Right Process

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4702168)