论文信息 - The Equivalence of Generalized Least Squares and Maximum Likelihood Estimates in the Exponential Family

The Equivalence of Generalized Least Squares and Maximum Likelihood Estimates in the Exponential Family

Abstract The method of iterative weighted least squares can be used to estimate the parameters in a nonlinear regression model. If the dependent variables are observations from a member of the regular exponential family, then under mild conditions it is shown that the IWLS estimates are identical to those obtained using the maximum likelihood principle. An application is provided to illustrate the results.

[1] P. McCullagh,et al. Generalized Linear Models , 1972, Predictive Analytics.

[2] Edwin L. Bradley,et al. The Equivalence of Maximum Likelihood and Weighted Least Squares Estimates in the Exponential Family , 1973 .

[3] Edward L. Frome,et al. Regression Analysis of Poisson-Distributed Data , 1973 .

[4] Olvi L. Mangasarian,et al. Nonlinear Programming , 1969 .

[5] Adi Ben-Israel,et al. Generalized inverses: theory and applications , 1974 .

[6] R H Moore,et al. The Use of Non-Linear Regression Methods for Analysing Sensitivity and Quantal Response Data , 1967 .

[7] S. Erlander,et al. On Asymptotic Simultaneous Confidence Regions for Regression Planes in a Poisson Model , 1972 .

[8] Malcolm E. Turner,et al. GENERALIZED ASYMPTOTIC REGRESSION AND NON-LINEAR PATH ANALYSIS1 , 1961 .