In this paper we use the significance of the attribute in rough set theory as the index to select splitting attributes for constructing the decision tree,and put forward a new decision tree classification algorithm S_D_Tree, of which the time complexity for selecting splitting attribute is O( ｜C｜ ｜n｜).Experimental results on three data sets demonstrate that the proposed algorithm can construct a less complex decision tree, and can also obtain comparative classification accuracy compared with C4.5.%采用粗糙集理论中的属性重要度作为挑选测试属性的指标来构造决策树,形成了一种新的决策树分类算法S_D_Tree,在计算挑选测试属性的时间复杂度为O(|C||U|).实验结果表明,该算法可以构建一个较简洁的决策树,与C4.5算法相比较,具有更好的预测准确率.
展开▼
