A new general approach to minimum distance estimation. (English) Zbl 0552.62016

Vajda, Igor

Information theory, statistical decision functions, random processes, Trans. 9th Prague Conf., Prague 1982, Vol. A, 103-112 (1983).

[For the entire collection see Zbl 0531.00015.]
The problem of minimum distance estimation of parametric statistical models is concerned. A general class of estimators based on the f- divergence by Csiszár is introduced. The theory of f-divergence estimators is shown to cover the traditional minimum distance estimators as particular cases.
The theory can also reveal the relation between the minimum distance estimators and the ML-estimators in general. The f-divergence estimators are shown to be regular enough such that their properties can be studied analytically.

Reviewer: I.Virtanen

MSC:

62F10

Point estimation

Keywords:

minimum divergence estimators; consistency; equivariance; efficiency; robustness; minimum distance estimation; f-divergence estimators; ML- estimators

Citations:

Zbl 0531.00015