满足二维空间邻接条件的遗传聚类算法研究

分析了目前满足二维空间邻接条件聚类算法的不足。从聚类概念出发，重新定义了满足二维空间邻接条件聚类的概念。面对满足二维空间邻接条件的聚类问题，定义了邻接矩阵的概念。以邻近距离和邻接矩阵为样本划分依据，以类内平方误差和（WGSS）为聚类目标函数，引入遗传算法，提出满足二维空间邻接条件的遗传聚类算法。通过实例进行了算法测试,并与模糊聚类（FCM）结果进行比较。

Research of the genetic-clustering algorithm considering the condition of planar adjacency relationship

The shortcomings about these days clustering algorithm considering the condition of planar adjacency relationship are analysised.The clustering algorithm considering the condition of planar adjacency relationship is defined again newly from the general clustering.In order to dealing with the clustering considering the condition of planar adjacency relationship,the concept adjacency matrix is defined.The genetic-clustering algorithm considering the condition of planar adjacency relationship is put forward,partitioning samples on the best-close distance and adjacency matrix,calculating cluster aim function on within-group sum of squares(WGSS) error,importing genetic algorithm.The algorithm is validated and compared with the FCM clustering outcome by examples.Algorithm testing show: the genetic-clustering algorithm considering the condition of planar adjacency relationship is completely feasible and availability.