基于用户兴趣度和MID3决策树改进方法

Improved decision tree algorithm based on user-interest and MID3

决策树是数据挖掘的重要方法,通常用来形成分类器和预测模型。分析ID3算法和现有的ID3改进算法所存在的问题。提出一种合理且可靠的MID3的改进算法,即针对MID3算法倾向于取值较多属性的缺点,引入了用户兴趣度对算法进行改进,改进后的MID3算法既可以在一定程度上解决多值偏向问题,也可以考虑决策树的两层节点。并通过实验与其他ID3改进算法进行比较,实验结果表明,改进后的MID3算法是有效的。

Decision tree is an important approach in data mining,which can be used to form classification and predictive model.In the present essay,ID3 algorithm and its improved algorithms at present are analyzed,and then a more reliable and reasonable approach is introduced in which user-interest is involved to solve the problem of multi-value and the double nodes as well.The improved MID 3 algorithm is proved efficient by the experiment in which several ID3 algorithms are compared.