混合值不完备信息系统一种新的数据分析方法

摘要

针对混合值不完备信息系统的完备化问题,提出了一种改进的ROUSTIDA算法.对可辨识矩阵的定义在邻域的范围内进行了扩充,使之可以适用于混合值的不完备信息系统,结合极大相容块技术给出了具体的改进算法,使填充能力得到很大的改善,同时还具备了初步排除噪声数据的能力.实例分析表明,该算法是行之有效的.%Thie paper proposed a new improved ROUSTIDA algorithm to deal with incomplete information system with hybrid value. Firstly, gave an extended definition of discernibility matrix to deal with incomplete information system with hybrid value within the scope of neighborhood, then expounded the specific method combined with the technique of maximal consistent block, so greatly improved the ability to pack data, meanwhile the method had the good capacity of eliminating the elementary noise. Finally, an example is given to show that the presented algorithm is reasonable and efficient.