一种基于多关系的朴素贝叶斯空间分类算法

空间分类既要考虑待分类对象的非空间属性,还要考虑其空间邻接对象非空间属性对分类的影响.提出一种基于多关系的朴素贝叶斯空间分类算法,算法将多关系分类方法用于空间分类,考虑了不同近邻对象的非空间属性对分类产生的影响,其分类准确率高于单关系朴素贝叶斯空间分类算法.算法可以用于空间数据库中的大数据集,不需要复杂的数据预处理.

One Naive Bayesian classification algorithm based on multi-relation data

When a spatial object being classified,not only its non-spatial attributes but also the non-spatial attributes of its spatial neighbor should be considered. This paper has proposed a spatial naive Bayesian classification algorithm based on multi-relational data. This algorithm utilizes the multi-relation classification method in spatial classification,and considers the influence on the classification result about non-spatial attributes of each neighboring object. The classification accuracy rate of this algorithm is better than that of the naive Bayesian classification algorithm based on single relation. This algorithm can carry out big data processing in the spatial database and does not require the intricate data pretreatment.