Improved LEM2 rule induction algorithm
Xu Yi, LI Long-shu, Xuejun Li · Systems Engineering - Theory & Practice · 2010
In view of the limitations of LEM2 algorithm,to improve the quality and efficiency of rule induction,improved LEM2 rule induction algorithm is proposed.Based on generalized decision,minimalrule -first strategy is given,which can make rule more laconic and improve the quality of rule induction. Redundancy attribute-value pair set is defined,by deleting the redundancy attribute-value pair from the candidate attribute-value pair sets,which can accelerate convergence rate and improve the efficiency of rule induction.The experimental results show that the improved LEM2 algorithm is more effective for rule induction from incomplete information system.