CRMMS: An Algorithm of Classification Rule Mining with Multiple Support Demand

The paper presents an algorithm CRMMS of classification rule mining with multi-support demand, which adopts the frequent classification item-set tree FCIST to organize the frequent pattern sets, and builds the array-based threaded transaction forest ATTF, and applies multiple supports for classification rules mining. The CRMMS uses the breath first strategy assisted by the depth first strategy, and adopts pseudo projection, which makes it unnecessary to scan the database, and to construct the projected transaction subsets repeatedly. The algorithm reduces the memory and time cost, and makes the projection more efficient and scalable. The CRMMS algorithm can be used in the consumerpsilas basket analysis, consumption behavior rules mining in the retailing industry.