Modified Power Divergence Estimators and Their Performances in Normal Models

Result code in IS VaVaI
<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21340%2F10%3A00176191" target="_blank" >RIV/68407700:21340/10:00176191 - isvavai.cz</a>
Result on the web
—
DOI - Digital Object Identifier
—

Result language
angličtina
Original language name
Modified Power Divergence Estimators and Their Performances in Normal Models
Original language description
As was already mentioned in many publications, the well known information-theoretic measures of divergence of probability measures cannot be directly applied in statistical estimation, since the divergence between the theoretical absolutely continuous probability measure and the discrete empirical probability measure is always equal to its upper bound and often takes on infinite values. In 2008 - 2009 Broniatowski & Vajda studied and extended two different modifications of divergences proposed independently in 2006. They altered the traditional phi-divergences into subdivergences and superdivergences and defined maximum subdivergence estimators with escort parameter theta and minimum superdivergence estimators. We shall present the key results of extensive simulation study of these types of point estimators. The main interest of our research was to examine these modifications in practical use as to the consistency, robustness and efficiency of the estimators.
Czech name
—
Czech description
—

Publication year
2010
Confidentiality
S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Similar results(10)