摘要： 在不完备决策表中，引入基于相容关系的相对知识量，定义属性的重要度，提出一种启发式属性约简算法。该算法通过在约简过程中不断缩小样本空间的方法，降低算法计算时间。采用UCI数据集进行实验，结果表明该算法可提高不完备决策表的约简效率，适用于大规模数据集的属性约简。
关键词:
粗糙集,
相对知识量,
不完备决策表,
属性约简,
相容关系,
属性重要度
Abstract: Relative knowledge quantity is defined under incomplete decision table, and the definition of the important degree of attributes is obtained. A heuristic algorithm based on relative knowledge quantity for reduction of attributes is presented. This algorithm reduces the time consumption through reducing the scale of data. Experiment on the UCI data set demonstrates the improvements of the reduction efficiency, especially for the data sets with large scale.
Key words:
rough set,
relative knowledge quantity,
incomplete decision table,
attribute reduction,
tolerance relation,
attribute significance
中图分类号:
韩晓琴, 孙士保, 张瑞玲. 不完备决策表中基于相对知识量的属性约简[J]. 计算机工程, 2012, 38(11): 59-61,65.
HAN Xiao-Qin, SUN Shi-Bao, ZHANG Rui-Ling. Attribute Reduction Based on Relative Knowledge Quantity in Incomplete Decision Table[J]. Computer Engineering, 2012, 38(11): 59-61,65.