决策树技术及其当前研究方向

Decision Tree Technique and its Current Research

介绍了决策树归纳技术及其发展过程,阐述了目前最流行的决策树技术的基本方法及简化决策树的主要方法。同时介绍了决策树技术面临的挑战,目前在与其他技术结合,寻找新的构造方法,简化方法,训练与检验数据的大小及特性与其本身特性的关系,不确定环境下决策,时间复杂度与准确性之间的关系,软件实现等方面的研究,以及它在工程上的应用,展望了它的未来发展前景。

An introduction and overview of decision tree induction technique is provided. The basic methods of the current technique of decision tree pruning are discussed. The key ways of simplifying decision trees are also discussed. The research directions of decision tree technique and its engineering applications are reviewed, such as integration with other techniques, better methods to construct and simplify, the impact of data size and quality on decision tree performances, how to build decision trees in an uncertain environment, the trade-off between time complexity and accuracy, software implementation of decision tree induction, etc. The future picture for decision tree techniques is given.