Score Function of Distribution and Revival of the Moment Method

Autoři: RNDr. Zdeněk Fabián, CSc.
Rok vydání: 2016
Typ publikace: Článek v odborném periodiku
Periodikum: Communications in Statistics - Theory and Methods
Projekty: Využití pokročilé statistické analýzy a nestatistických separačních metod pro detekování fyzikálních procesů v datech snímaných urychlovači elementárních částic.
Bibliografická citace: FABIÁN, Zdeněk. Score Function of Distribution and Revival of the Moment Method. Communications in Statistics - Theory and Methods. 2016, 45(4), 1118-1136. ISSN 0361-0926. E-ISSN 1532-415X
Dostupné z: doi:10.1080/03610926.2013.857688
DOI: 10.1080/03610926.2013.857688
ID: 458402

Abstrakt

The article deals with the scalar-valued score function defined for any regular unimodal and continuous probability distribution with arbitrary interval support, recently introduced by the author. Function values describe the relative influence of observed values under the given model with respect to a certain 'center point' of the distribution. By means of the function is introduced a generalized moment method of parametric estimation, giving estimates with a 'natural' tradeoff between efficiency a robustness. Inference function used in the estimating equations is either the bounded score function of distribution of heavy-tailed models or its simple modification based on the Huber’s approach if the function is unbounded.