Back to Search
Start Over
Efficient and superefficient estimators of filtered Poisson process intensities.
- Source :
- Communications in Statistics: Theory & Methods; 2019, Vol. 48 Issue 7, p1682-1692, 11p
- Publication Year :
- 2019
-
Abstract
- Let N<superscript>K</superscript> = {N<superscript>K</superscript><subscript>t</subscript>, t ∈ [0, T]} be a filtered Poisson process defined on a probability space , and let θ ≔ (θ<subscript>t</subscript>, t ∈ [0, T]) be a deterministic function which is the intensity of N<superscript>K</superscript> under a probability P<subscript>θ</subscript>. In the present paper we prove that the natural maximum likelihood estimator (MLE) N<superscript>K</superscript> is an efficient estimator for θ under P<subscript>θ</subscript>. Using Malliavin calculus we construct superefficient estimators of Stein type for θ which dominate, under the usual quadratic risk, the MLE N<superscript>K</superscript>. These superefficient estimators are given under the form where F is a random variable satisfying some assumptions and is the Malliavin derivative with respect to the compensated version of N<superscript>K</superscript>. [ABSTRACT FROM AUTHOR]
- Subjects :
- POISSON processes
MALLIAVIN calculus
RANDOM variables
PROBABILITY theory
Subjects
Details
- Language :
- English
- ISSN :
- 03610926
- Volume :
- 48
- Issue :
- 7
- Database :
- Complementary Index
- Journal :
- Communications in Statistics: Theory & Methods
- Publication Type :
- Academic Journal
- Accession number :
- 137165989
- Full Text :
- https://doi.org/10.1080/03610926.2018.1438622