关联分类算法改进及其在中药“性-效”分析中的应用

Improvement of Association Classification Algorithm and Its Application in“Performance-Efficacy”Analysis of Traditional Chinese Medicine

针对中药“性-效”数据关联度高、属性稀疏的问题,提出一种使用垂直数据格式生成类关联规则的关联分类算法(ECBA)。该算法通过将数据转换为垂直格式而避免了经典关联分类算法(CBA)生成大量候选规则集、频繁遍历数据库、产生无意义分类规则等不足。实验结果表明,相较于传统算法CBA,改进算法ECBA在规则生成时间、规则有效性以及准确率方面均有明显提升,更适用于中药“性-效”数据分析。

Aiming at the problem of high correlation and sparse attributes of traditional Chinese medicine“performance-efficacy”data,an Association Classification Algorithm(ECBA)for generating class association rules using vertical data format is proposed.This algorithm avoids the shortcomings of the classical association classification algorithm(CBA)in generating a large number of candidate rule sets,frequently traversing the database,and generating meaningless classification rules by converting the data into a vertical format.The experimental results show that compared to the traditional algorithm CBA,the improved algorithm ECBA has significantly improved in rule generation time,rule effectiveness,and accuracy,making it more suitable for analyzing the“performance effectiveness”data of traditional Chinese medicine.