急切式和懒惰式学习策略相结合的决策树分类模型

A Decision-Tree Classifier Hybrid Model of Eager Strategy and Lazy Strategy

急切式学习策略和懒惰式学习策略有着不同的学习和分类机制.通过分析急切式学习策略下的普通决策树模型和懒惰式学习策略下的懒惰式决策树模型,提出了一种新的决策树分类模型即Semi-LDtree.它生成的决策树的结点,如普通决策树一样,包含单变量分裂,但是叶子结点相当于一个懒惰式决策树分类器.这种分类模型保留了普通决策树良好的可解释性,实验结果表明它提高了分类速度和分类精确度,特别是在大的数据集合上效果更加明显.

The eager strategy and lazy strategy have different learning and classification mechanism. On the basis of analyzing regular decision tree classification model adopting eager strategy and lazy decision-tree classification model adopting lazy strategy, we propose a new decision-tree classification model, Semi-LDtree： the decision-tree nodes contain univariate splits as regular decision-trees, but the leaves contain lazy decision-tree classifiers. This classification model retains the good interpretability of decision-tree. The experimental results show this model has the higher classification accuracy and faster speed, especially on the larger databases tested.