摘要: 针对属性在不断增加的决策表,为了快速准确地计算出属性约简,提出一种增量式属性约简算法。以正域为约简的标准,利用贪心算法思想,以属性区分能力为选择标准,逐渐构造近似的属性约简,从中删减掉不必要的属性,最终得到属性约简。经复杂度分析与实验数据测试,证明该算法的复杂度低并且约简结果准确。
关键词:
粗糙集,
改进区分矩阵,
增量式属性约简,
决策表,
贪心算法
Abstract: In order to efficiently reduce attributes of tables with incremental attributes, an incremental attributes reduction is proposed. It figures out the incremental attributes reduction with greedy algorithms, which increasingly constructs approximate reduction and then kickes out unnecessary attributes. The algorithm is valid according to complex analysis and data tests.
Key words:
rough set,
improved discernibility matrix,
incremental attributes reduction,
decision table,
greedy algorithm
中图分类号:
刘高峰, 牟廉明, 张涛. 基于改进区分矩阵的决策表增量式属性约简[J]. 计算机工程, 2010, 36(20): 46-48.
LIU Gao-Feng, MAO Lian-Meng, ZHANG Chao. Incremental Attributes Reduction of Decision Table Based on Improved Discernibility Matrix[J]. Computer Engineering, 2010, 36(20): 46-48.