Maximum likelihood estimation of probabilistic choice methods
Maximum likelihood estimation of probabilistic choice methods
复制标题
概率选择方法的最大似然估计
DOI：
--
复制
发表时间：
1987
期刊：
影响因子：
--
通讯作者：
D. S. Bunch
中科院分区：
文献类型：
--
作者：
D. S. Bunch
Probabilistic choice models are used by social scientists in the analysis of individual choice behavior involving discrete choice alternatives. The development of new complex choice models has given rise to a need for algorithms and software for parameter estimation of choice models. One important estimator is the maximum likelihood estimator, which historically has been avoided due to the necessity of solving a nonlinear optimization problem. New algorithms for maximum likelihood estimation of a general probabilistic choice model that exploit special structure are developed; the MLE problem for choice models can be written as a problem in generalized regression. The algorithms are implemented using state-of-the-art nonlinear optimization techniques including model/trust regions, least-change secant updates, and model switching. The algorithms perform well in tests on the multinomial logic model, the elimination-by-aspects model, and a new model of market share by Batsell and Polking.
