A Simple Identification Proof for a Mixture of Two Univariate Normal Distributions |
| |
Authors: | Erik Meijer Jelmer Y Ypma |
| |
Institution: | (1) RAND Corporation and University of Groningen, Santa Monica, USA;(2) University College London and University of Groningen, Santa Monica, USA;(3) RAND Corporation, 1776 Main St, PO Box 2138, Santa Monica, CA 90407-2138, USA |
| |
Abstract: | A simple proof of the identification of a mixture of two univariate normal distributions is given. The proof is based on the
equivalence of local identification with positive definiteness of the information matrix and the equivalence of the latter
to a condition on the score vector that is easily checked for this model. Two extensions using the same line of proof are
also given.
We would like to thank Tom Wansbeek, Michel Wedel, Arie Kapteyn, and two anonymous reviewers for helpful comments on earlier
versions of this paper. |
| |
Keywords: | Finite mixtures Information matrix Exponential family Mixture regression |
本文献已被 SpringerLink 等数据库收录! |
|