改进的Apriori算法在试题关联分析中的应用

针对关联规则挖掘中经典Apriori算法由于多次扫描数据、产生大量候选集及产生候选集时连接次数多等缺陷,导致效率较低。文中提出删除部分特殊事务,减少扫描数据次数。在生成候选k-项集前,对频繁k-1项集进行约简,减少连接次数和候选k-项集数,对Apriori算法进行改进。并将改进的Apriori算法用于试题分析中,得出试题之间的关联关系。实例表明,改进后的算法在效率上优于Apriori算法。

An Improved Apriori Algorithm with Application in Association Analysis of Examination

As a classical association rule mining algorithm, Apriori scans the database repeatedly and has a large amount of connection times before the generation of candidate item sets, and a huge number of candidate item sets are generated. These defects result in low efficiency. An improved algorithm is proposed, which works in the following two steps. Firstly, some special affairs are deleted to reduce the times of scanning data. Secondly, the frequent k-1 item sets are reduced before generating candidate k-item sets, so the times of the frequent k-1 item sets connection and the number of candidate k item sets is reduced. Finally the improved Apriori algorithm is used in the analysis of the item relation of an examination to find the relationship among various questions. Experiments show that the improved algorithm is superior to Apriori algorithm in efficiency, and can get a good result.