A Clustering Algorithm Basedon Discretized Interval Value
Shao Liang-shan · 2006
Inordertoimprove thequality oftraditional clustering algorithm andprevent thedistribution ofdata fromaffecting theclustering algorithm greatly, aclustering algorithm basedoninterval value wasproposed. Depending on theconsistency ofcondition attributes anddecision attributes inthedecision table, thedatawasdiscretized and attributes werereducedby usingdatasuper-cube and information entropy. Basedontheabove, thealgorithm can usetheadditivity ofsetfeature vector tocluster datajust by scanning thedecision tableonlyonetime.Experimental results indicate thatthealgorithm isefficient andeffective.