Abderrahim Siam2024-03-132024-03-132021http://dspace.univ-khenchela.dz:4000/handle/123456789/4239Associationrulesarethespecificdataminingmethodsaimingtodiscoverexplicitrelationsbetweenthe differentattributesinalargedataset.However,inreality,severaldatasetsmaycontainbothnumericand categoricalattributes.Recently,manymeta-heuristicalgorithmsthatmimicthenaturearedeveloped forsolvingcontinuousproblems.Thisarticleproposesanewalgorithm,DCSA-QAR,formining quantitativeassociationrulesbasedoncrowsearchalgorithm(CSA).Toaccomplishthis,newoperators aredefinedtoincreasetheabilitytoexplorethesearchingspaceandensurethetransitionfromthe continuoustothediscreteversionofCSA.Moreover,anewdiscretizationalgorithmisadoptedfor numericalattributestakingintoaccountdependenciesprobablythatexistbetweenattributes.Finally, toevaluatetheperformance,DCSA-QARiscomparedwithparticleswarmoptimizationandmono andmulti-objectiveevolutionaryapproachesforminingassociationrules.Theresultsobtainedover real-worlddatasetsshowtheoutstandingperformanceofDCSA-QARintermsofqualitymeasures.en