论文部分内容阅读
目前粗糙集模型属性约简大多是基于静态信息系统,而实际决策表的数据信息都是动态变化的,为了有效地对这些数据集进行属性约简处理,介绍了关系矩阵增量机制,提出一种基于关系矩阵的增量式属性约简算法,在原有等价关系矩阵和约简的基础上,当决策表增加了一些对象,对决策表的等价关系矩阵和属性约简进行更新,便能快速求解出更新后的决策表属性约简.最后通过实例分析以及在UCI的2个数据集上分别对增量和非增量式的方法的性能进行了测试,并将实验结果进行比较,结果表明了增量式约简算法的有效性和正确性.
At present, rough set model attribute reduction is mostly based on static information system, but the actual decision table data information are dynamically changing. In order to effectively reduce the attribute set of these data sets, this paper introduces the incremental mechanism of relation matrix, Based on the relation matrix-based incremental attribute reduction algorithm, based on the original equivalent relation matrix and reduction, when some objects are added to the decision table and the equivalence relation matrix and attribute reduction of the decision table are updated, The updated decision table attribute reduction was quickly solved.Finally, the performance of incremental and non-incremental methods were tested respectively on two data sets of UCI through case analysis and comparison, and the results were compared It shows the validity and correctness of incremental reduction algorithm.