基于时间序列模型的动态关联规则元规则挖掘

摘要

Almost all of association rules mining algorithms are considered as static ones. In fact, it is possible that a rule will change greatly with time, this paper introduces a rule called dynamic association rule and the definition of the meta-asseciation rules for dynamic association role, focuses on introducing the method which by the model of time series to mine the meta-association rules for dynamic association rule, and proves that this method is fit on the historical data. It can establish the equation to forecast the tendency of changing rules and mine the meta-association roles for dynamic association rule.%针对现有关联规则挖掘算法大多是挖掘一种静态关联规则的情况,介绍动态关联规则的定义,给出动态关联规则元规则的形式化定义,解决规则随时间的推移可能会有很大变化的情况下为规则建立元规则的问题,描述一种基于时间序列模型的预测和分析动态关联规则的元规则的方法,从而较好地拟合历史数据,给出满足一定显著性水平预测趋势模型的方程,挖掘规则的变化趋势,为规则建立元规则.