A new algorithm based on an FC-tree (frequent closed pattern tree) and a max-FCIA (maximal frequent closed itemsets algorithm) is presented, which is used to mine the frequent closed itemsets for solving memory an...A new algorithm based on an FC-tree (frequent closed pattern tree) and a max-FCIA (maximal frequent closed itemsets algorithm) is presented, which is used to mine the frequent closed itemsets for solving memory and time consuming problems. This algorithm maps the transaction database by using a Hash table,gets the support of all frequent itemsets through operating the Hash table and forms a lexicographic subset tree including the frequent itemsets.Efficient pruning methods are used to get the FC-tree including all the minimum frequent closed itemsets through processing the lexicographic subset tree.Finally,frequent closed itemsets are generated from minimum frequent closed itemsets.The experimental results show that the mapping transaction database is introduced in the algorithm to reduce time consumption and to improve the efficiency of the program.Furthermore,the effective pruning strategy restrains the number of candidates,which saves space.The results show that the algorithm is effective.展开更多
针对传统多层关联分类挖掘产生大量冗余规则而影响分类效率的问题,提出了一种基于类FP-tree的多层关联分类器MACCF(Multi-level Associative Classifier based on Class FP-tree)。该分类器依据事务的类标号划分训练集,采用闭频繁模式(C...针对传统多层关联分类挖掘产生大量冗余规则而影响分类效率的问题,提出了一种基于类FP-tree的多层关联分类器MACCF(Multi-level Associative Classifier based on Class FP-tree)。该分类器依据事务的类标号划分训练集,采用闭频繁模式(CLOSET+)产生完全候选项目集,通过设计适当的类内规则剪枝策略和类间规则剪枝策略,减少了大量冗余的分类规则,提高了分类的准确率;采用交叉关联规则方法,解决了交叉层数据的分类问题,实验结果表明了算法的高效性。展开更多
基金The National Natural Science Foundation of China(No.60603047)the Natural Science Foundation of Liaoning ProvinceLiaoning Higher Education Research Foundation(No.2008341)
文摘A new algorithm based on an FC-tree (frequent closed pattern tree) and a max-FCIA (maximal frequent closed itemsets algorithm) is presented, which is used to mine the frequent closed itemsets for solving memory and time consuming problems. This algorithm maps the transaction database by using a Hash table,gets the support of all frequent itemsets through operating the Hash table and forms a lexicographic subset tree including the frequent itemsets.Efficient pruning methods are used to get the FC-tree including all the minimum frequent closed itemsets through processing the lexicographic subset tree.Finally,frequent closed itemsets are generated from minimum frequent closed itemsets.The experimental results show that the mapping transaction database is introduced in the algorithm to reduce time consumption and to improve the efficiency of the program.Furthermore,the effective pruning strategy restrains the number of candidates,which saves space.The results show that the algorithm is effective.
文摘针对传统多层关联分类挖掘产生大量冗余规则而影响分类效率的问题,提出了一种基于类FP-tree的多层关联分类器MACCF(Multi-level Associative Classifier based on Class FP-tree)。该分类器依据事务的类标号划分训练集,采用闭频繁模式(CLOSET+)产生完全候选项目集,通过设计适当的类内规则剪枝策略和类间规则剪枝策略,减少了大量冗余的分类规则,提高了分类的准确率;采用交叉关联规则方法,解决了交叉层数据的分类问题,实验结果表明了算法的高效性。