基于GM(1,1)模型的元规则挖掘研究

Research of Mining the Meta-association Rules Based on the Model of GM(1.1)

关联规则挖掘主要用于发现事务数据集中项与项之间的关系,由于事务数据通常具有时间特性,同一规则在不同的时间段,其支持度和置信度值也不尽相同。为关联规则建立元规则,对其支持度和置信度变化趋势进行分析和预测,有利于进一步指导挖掘和决策。本文通过一个例子,分析了使用GM(1,1)模型进行元规则挖掘的一般过程,评价了GM(1,1)模型在元规则挖掘中的优缺点。

The association rule mining aims at the relationships between the items of the transaction data sets. Because the transaction data are assigned to different sets according to their time stamps, the Support value and the Confidence value of a certain rule is not quite the same at each set. By analyzing the changes of a rule, the Meta-association Rules will be found and be used to predict the next Support value or/and the next Confidence value of it. The established Meta-association Rules can also be used to direct the further mining operations and the decisions. With an example, this paper analysis the process of mining the Meta-association Rules by the model of GM（1.1）. By comparing the result with the result of fitting polynomial to the rule, the paper discusses the advantages and disadvantages of using the model of GM（1.1） to mine the Meta-association Rules.