基于项目排序和权值矩阵的关联规则挖掘算法

Association Rules Mining Algorithm Based on Item Ranking and Weight Matrix

针对当前加权关联规则算法中存在的运算效率低、无关规则量大、有效规则难以发现的问题,提出一种基于项目排序和权值矩阵的关联规则挖掘算法。该算法首先将各项目依关注度进行排序,并根据排名给每个项目分配权重;接着给出新的项集权值计算模型,针对该模型重新定义项集加权支持度和加权置信度,并通过权值矩阵的构造、候选集的剪枝操作来达到快速挖掘项集规则的目的;最后通过设计对比实验,验证算法的性能及挖掘结果的有效性。实验表明,该方法不仅压缩了频繁集和规则数量,提升了算法效率,而且保留了高关注度的项集规则,并能将这些规则排在规则集顶层,方便决策者快速发现感兴趣的规则。

To address low operation efficiency,large amount of irrelevant rules and difficult discovery of effective rules in current weighted association rules algorithm,an algorithm for mining association rules based on item ranking and weight matrix is proposed.First,the items are ranked according to their concern and weights are assigned to each item according to the ranking.Then a new itemset weight calculation model is proposed,and the weighted support and confidence of itemsets are redefined for this model.Through the construction of weight matrix and pruning operation of candidate set,the purpose of mining item rules is quickly achieved.Finally,the performance of the algorithm and the effectiveness of the mining results are verified through the design comparison experiments.Experiments show that this method not only reduces the number of frequent sets and rules,improves the efficiency of the algorithm,but also retains the itemset rules with high concern,and can arrange these rules at the top level of the rule set,so as to facilitate quick discovering of the rules of interest for the decision makers.