Abstract: Aiming at the low efficiency problem of the construction of decision tree in large-scale training units, this paper presents an improved construction method for decision tree. This method uses FP_Tree algorithm to save more memory than Apriori algorithm. It takes the place of split algorithm of the training units in classical algorithm by the path of FP_Tree, and gets the same decision tree model as the original algorithm. Test results show that the improved method has good property.

Key words: decision tree, FP_Tree algorithm, classification

摘要： 针对大规模训练元组决策树构造效率较低的问题，提出一种改进的决策树构造方法。该方法利用FP_Tree算法，比采用经典Apriori算法节省了更多内存开销。使用FP_Tree路径替代经典算法中训练元组的分裂计算，得到与原算法相同的决策树模型。实验结果证明，改进后的方法具有良好性能。

关键词: 决策树, FP_Tree算法, 分类

CLC Number: 

• TP391