摘要
Bagging通过组合不稳定的分类器在很大程度上降低了"弱"学习算法的分类误差。基于Torsten等人提出的Double-Bagging算法,本文对其加以修改并应用于基因微阵列数据的处理。在给定的训练数据集和测试集上试验并比较了多种分类器,结果表明Double-Bagging决策树分类精确度优于Bagging决策树和C4.5算法。
The three algorithms: Bagging, Double-Bagging and C4.5 were introduced, of which Dou ble-Bagging algorithm was modified and the modified algorithm was applied to gene micro-array data processing. The experiment was done on the specific training data set and the test collection, and many kinds of sorters were compared. The results show that the classification precision of the Dou ble-Bagging decision-making tree is superior to the Bagging decision-making tree and C4.5.
出处
《湖北汽车工业学院学报》
2009年第2期40-43,共4页
Journal of Hubei University Of Automotive Technology