摘要
许多属性约简算法建立在整个决策表上,当数据集较大时,效率往往较低。针对这一问题,提出一种基于决策表分解的属性约简算法。结合正域基数和互信息增益来衡量属性的重要性,其求解策略是基于每次迭代后所分解的决策表计算正域基数和互信息增益,以不断减少对象集,降低计算时间。进行时间复杂度分析,通过实验对该算法的效率进行测试,测试结果表明,该算法是正确﹑高效的。
Many existing algorithms of attributes reduction based on the entire decision table. When the data set is large, the efficiency of these algorithms is often low. To overcome this shortcoming, a new attribute reduction algorithm based on decision table decomposition was proposed. The algorithm combined the positive region and the gain of mutual information to measure the importance of an attribute. The reduction strategy of the algorithm was to compute the positive region and the mutual information gain based on the decomposed decision table in the next iteration. By reducing the set of objects, the time complexity was decreased. Time complexity of the algorithm was analyzed and the efficiency of that was tested. Experimental results show that the algorithm is correct and efficient.
出处
《计算机工程与设计》
CSCD
北大核心
2014年第8期2872-2875,2943,共5页
Computer Engineering and Design
关键词
属性约简
粗糙集理论
正域基数
互信息增益
决策表分解
attribute reduction
rough set theory
positive region
mutual information gain
decision table decomposition