基于基因表达式编程的ISODATA模糊聚类算法

针对ISODATA算法需要人为给定分类数，对初始聚类中心较为敏感，没有显示出自动聚类效果等不足，结合基因表达式编程(GEP)嵌套构成迭代自组织模糊聚类进行优化计算。该方法不仅能在不需要先验知识的条件下对数据进行自动聚类，而且充分利用了GEP算法的全局寻优能力及ISODATA算法的软性分类特性，提高了算法的收敛速度和聚类精度。通过仿真验证及对比分析，运用到地理信息系统(GIS)物流选址实际问题中，得到了理想聚类效果。

Fuzzy ISODATA clustering algorithm based on gene expression programming

Concerning the defects of the artificial setting of the categories number, the sensitiveness to initial cluster centers and the lack of automatic clustering effects on the ISODATA algorithm, in combination with Gene Expression Programming (GEP) a nested iterative self organizing fuzzy clustering was formed up. This paper presented a new algorithm: fuzzy ISODATA clustering algorithm based on GEP. This algorithm not only conducted automatic clustering under the condition of no prior knowledge, but also fully used the capability of global optimization of GEP algorithm and soft classification features of ISODATA, which resulted in the increase of the convergence speed and the clustering accuracy. It is verified by simulation and comparative analysis of the practical problems in GIS logistics location.