Predicting Accident Frequencies for Drivers Classified by Two Factors
本文提出一系列泊松数据的对数线性模型，结合最大似然和经验贝叶斯估计，统一了风险分类、平滑、可信度理论和经验费率等精算概念，并用加州事故数据评估方法表现。
Abstract For predicting accident frequencies, a succession of log-linear models for Poisson data, some of which include nested random effects, is introduced. By applying maximum likelihood and empirical Bayes estimation techniques to these models, one can incorporate the actuarial notions of risk classification, model-based smoothing, credibility theory, and experience rating under a unified statistical approach to loss prediction. The performance of these methods is evaluated by using accident data from California.