A Novel Classification Algorithm Based on Association Rules Mining

摘要: The traditional methods for mining classification rules such as heuristics or greedy only generate the that are too general overfitting to do with given database. Thus, they introduce high error ratio. Recently, a new method of is proposed: based on association (CARs). It more advantageous than in it removes noise and therefore accuracy higher. In this paper, we propose ECR-CARM algorithm. ECR-tree find all CARs. Besides that, necessary redundant pruning reducing gain smaller set (i.e., time identifying class cases increasing accuracy). We also develop property fast prune rules.