在费率厘定中,当索赔次数数据存在过离散(over-dispersion)特征时,通常会采用负二项回归模型,但当索赔数据中同时又出现零膨胀(zero-inflated)问题时,负二项回归模型不再适合对这样的数据进行分析.在传统的零膨胀负二项回归模型为基础,并将其推广到更为一般的形式,同时利用解决费率厘定中出现的既有过离散又有零膨胀的问题.通过对一组汽车损失数据的拟合,推广后的零膨胀负二项回归模型有效地改善了拟合效果. Negative binomial regression models are commonly used when there is over-dispersion of claims data in rate determination, but when zero-inflated problems appear in the claim data at the same time, the negative two Term regression model is no longer suitable for such data analysis.In the traditional zero-expansion negative binomial regression model based on, and its generalization to a more general form, at the same time using the solution to determine the rate of existing both discrete Another zero-expansion problem.Through the fitting of a set of automobile loss data, the negative zero-expansion negative binomial regression model can effectively improve the fitting effect.