Abstract: A menu-based choice experimental survey is conducted in Wuxi, Jiangsu Province. An improved k-modes algorithm is used to analyze the data. The improved k-modes algorithm optimizes the method of selecting initial cluster centers so as to simplify the process of clustering. Cluster modes are replaced by the pattern of thinking all property values thereby improving the clustering accuracy. Experimental results indicate that customer can be divided into four classes, i. e. those who pay no attention to traceability information class, those who pay attention to farming information class, those who attach importance to the traceability information, and those who pay attention to farming information and government-certified class. Pork with different combinations of traceability information can be provided to different customer classes thus expanding consumers’ demand for traceable food and finally improving food safety standards.

Key words: clustering, menu based choice experiment, traceability pork, willingness to pay

摘要： 运用菜单法问卷的调查方式，以江苏省无锡市区消费者为调查对象，以可追溯猪肉为案例，基于改进的k-modes聚类方法，研究消费者对猪肉可追溯属性的支付意愿。改进的k-modes聚类方法优化初始聚类中心选取从而简化聚类过程，以考虑可追溯属性的所有属性值的模式代替聚类的modes，从而提高聚类精确性。实验结果表明，消费者可分为不重视可追溯信息类、重视养殖信息类、重视可追溯信息类、重视养殖信息及政府认证类四个群体。可针对不同的群体提供不同的可追溯属性的组合的猪肉以扩大消费者对可追溯食品的需求，提高食品安全保障水平。

关键词: 聚类, 菜单法, 可追溯猪肉, 支付意愿