基于粗集可辨识矩阵的属性频率约简算法

Attribute Frequency Reduction Arithmetic Based on Discernibile Matrix of Rough Set

针对信息系统在属性约简过程中存在属性频率值相同的问题进行改进,改进后的算法在基于可辨识矩阵属性频率约简算法的基础上,引进强等价集概念,以属性在可辨识矩阵中出现的次数越多其重要性越大为启发式信息,利用强等价集中的属性是可以约简的特性,在属性频率约简过程中判断具有相同属性频率属性是否最终包含在核属性集里,提出改进的属性频率约简算法。通过理论和实例的分析证明,该算法在保持时间复杂度不变的情况下,处理具有相同属性频率信息系统的属性约简,使其准确性得到提高,与原算法相比,改进后的算法可以得到一个更为精准的约简结果。

In the view of improving system exsists the same attribute frequency value in attribute reduction, the improved algorithm based on discernibile matrix affribute frequency reduction arithmentic, the concept of strong equivalent set is introduced to identify attributes in the matrix in the number of the more importance to the greater heuristics. Using strong focus on the strong equivalent set is the reduction of the frequency attribute reduction in judgements have the same attributes＇ frequency of property included in the final set of attributes, and the properties frequency reduction algorithm is improved. Through theoretical analysis and examples prove that the algorithm to maintain the same time complexity of the situation, to deal with the same attributes＇ frequency of information systems attribute reduction, the accuracy is improved,compared with the original algorithm,improved algorithm can get a more precise reduction results.