An approximate maximum likelihood procedure for parameter estimation in multivariate discrete data regression models |
| |
Authors: | Andrew W. Roddam |
| |
Affiliation: | a Department of Statistics, University of Oxford, UK. |
| |
Abstract: | This paper considers an alternative to iterative procedures used to calculate maximum likelihood estimates of regression coefficients in a general class of discrete data regression models. These models can include both marginal and conditional models and also local regression models. The classical estimation procedure is generally via a Fisher-scoring algorithm and can be computationally intensive for high-dimensional problems. The alternative method proposed here is non-iterative and is likely to be more efficient in high-dimensional problems. The method is demonstrated on two different classes of regression models. |
| |
Keywords: | |
本文献已被 InformaWorld 等数据库收录! |