多种属性选取标准“多数表决”优化决策树研究

Majority voting optimization of decision tree with multiple attributes selection criteria

测试属性的选取即属性选择标准是构建决策树的关键及核心,对于同样的数据集,不同的属性选取标准构建的决策树有可能差别很大。对于不知采用何种属性选择标准或者没有一种标准适合所处理的数据集,本文提出了一种解决的方法,即多种属性选取标准多数表决优化决策树算法,该算法利用"专家会诊"的思想,构建决策树,具有更广的适应性和更可能高的准确率。

The selection of test attributes is the key and core of constructing the decision tree.For the same data set,the decision trees constructed by different attribute selection criteria may vary greatly.Not to know which attribute selection criteria is good or not a standard suitable for handling data sets,this paper presents a solution,namely the majority voting optimization of decision tree with multiple attributes selection criteria algorithm,which uses the idea of“expert consultation”to construct the decision tree,with a wider adaptability and higher accuracy in the practical application.