中文摘要：

对非寿险产品分类费率的厘定通常采用单项分析法、最小偏差法和多元线性回归等方法。虽然这些方法在非寿险产品定价中仍然占有一度之地，但由于保险数据的特殊性，它们的缺陷越来越受到人们的重视。本文简要分析了这些传统定价方法存在的缺陷，介绍了非寿险精算中典型的广义线性模型，并通过汽车第三者责任保险的损失数据说明了广义线性模型在非寿险产品定价中的具体应用，以及应用广义线性模型时应该注意的几个问题。

英文摘要：

One-way analysis, minimum bias procedures and multiple regressions have been widely used in non-life classification rating. But their drawbacks are becoming very clear when considering properties of insurance data. The paper analyzes the drawbacks of these traditional methods, discusses the application of generalized linear models in non-life insurance pricing, and points out some problems to be considered when applying generalized linear models.