State Estimation for Cox Processes with Unknown Law: Parametric Models.
Abstract
Let (N sub i, M sub i) be i.i.d. copies of a Cox pair (N,M), with the Cox processes N sub i, but not the directing measures M Sub i, observable. Suppose that the probability law of (N,M) belongs to a finite-dimensional parametric family sub theta but is otherwise unknown. Approximations are derived for state estimators. Under standard smoothness and reqularity assumptions n times the difference between the ture and pseduo-stae estimators converges in distribution to a Gaussian random measure, with respect to the variation norm topology. Computation of maximum likelihood estimators by the EM algorithm is discussed in general and for two specific examples. Keywords: Parametric statistical model; Maximum likelihood estimator; EM algorithm.
Document Details
- Document Type
- Technical Report
- Publication Date
- Nov 01, 1985
- Accession Number
- ADA166179
Entities
People
- Alan F. Karr
Organizations
- Johns Hopkins University