基于模糊逻辑华莱士分类法的软件项目风险评估(IJIEEB-V5-N4-8)

I.J.InformationEngineeringandElectronicBusiness,2013,4,58-64PublishedOnlineOctober2013inMECS()DOI:10.5815/ijieeb.2013.04.08Copyright©2013MECSI.J.InformationEngineeringandElectronicBusiness,2013,4,58-64AssessmentofEffectiveRiskinSoftwareProjectsbasedonWallace’sClassificationUsingFuzzyLogicAliYavariMazandaranUniversityofSciencesandTechnology,IranEmail:[e-mail]nUniversityofSciencesandTechnology,IranEmail:[e-mail]uralSciencesandNaturalResourcesUniversityofGorgan,IranEmail:[e-mail]—Softwaredevelopmentalwaysfacesunexpectedeventssuchastechnologychanges,environmentalchanges,changinguserneeds.Thesechangeswillincreasetheriskinsoftwareprojects.Weneedtoriskmanagementtodealwithsoftwarerisks.Riskassessmentisoneofthemostimportantfactorsinriskandprojectmanagementofsoftwareprojects.Inthispaper,weuseWallace’sworkandfivefactorstopresentanefficientmethodtomeasuresoftwareriskusingfuzzylogic.Team,Planning,Complexity,RequirementsandUserarefactorsthatweuseinthispaper.Resultsofexperimentsshowsthatourframeworkismoreefficientthanotherframeworksandapproachesforriskassessmentinsoftwareprojects.IndexTerms—SoftwareRisk,Assessment,Wallace’sClassification,FuzzyLogic.I.INTRODUCTIONSoftwaredevelopmentprojectsfaceanumberofsoftwarerisks.Themostimportantfactorsthatmaycausefailureofprojectarerelatedtopoorperformance,teampressure,lowqualityandhighcost[1].Sosoftwareprojectriskmanagementplaysanimportantroleincompletingsoftwareprojectssuccessfully.Itconsistsofthefollowingfourphases:identification,assessment,planandcontrol.Riskassessmentisthebaseofsoftwareprojectriskmanagement.Accordingto[2]definition,riskexposureisexpressedastherelationshipRE=P(uo)*L(uo),whereREistheriskexposure,P(uo)istheprobabilityofanunsatisfactoryoutcomeandL(uo)isthelosstothepartiesaffectediftheoutcomeisunsatisfied[3].Themajorproblemassociatedwiththeestimationofrisksisthattheinputdataareimprecisebynatureanditisdifficulttorepresentthemwithcrispnumbers.Usuallytheriskanalystpreferstoestimateinlinguistictermssuchashighorlowratherthaninexactprobabilisticterminology.Tothisend,theapplicationofFuzzySetTheory(FST)toriskanalysisseemsappropriate;assuchanalysiscanhandlesubjectivityaswellasinexactandvagueinformation[1].Thesolutionthatissuggestedheretoovercomepreviouslymentionedproblemsistousefuzzylogiclinguisticvariablesforthecomplexitymetricsandmodel.Fuzzylogicisamathematicaltoolfordealingwithuncertaintiesandalsoitprovidesatechniquetodealwithimprecisionandinformationgranularity.Fuzzylogicisseenasameansofapproximatereasoning.Thefuzzylogichasbeensuccessfullyappliedinmanysegmentssuchasengineering,psychology,artificialintelligence,medicineandsociology.Inthispaperweproposeafuzzylogicapproachforriskestimationinsoftwareprojects.Weusefivefactorsthatareimportantinriskmanagement.ThosefactorsareTeam,Planning,Complexity,RequirementsandUser.Ourfuzzymodelhasfiveinputandweuse3linguisticvariablesforeachinput.ThismembershipfunctionsareLow,MediumandHigh.Fuzzificationmodulemapsthenon-fuzzyvaluesinfuzzyspaceanddefuzzificationmoduleconvertthefuzzynumbersintocrispspace.Inourframework,afterfuzzificaitionofeachinput,theMamdaniinferencesystemthatincludesarulebasecanevaluatetheriskofsoftwareproject.Thisisafuzzynumber,soitshouldconverttoacrispnumberbyadefuzzificaionalgorithm.Numberofrulesinthisrulebasethatuseinthissystemis243and“AndMethod“isminalso“OrMethod”ismax,thismeanthatandoperatorbetweentwonumbersselectminimumofnumbersandoroperatorbetweentwonumberselectmaximumofthisnumbers.Fuzzyinferenceistheactualprocessofmappingfromagiveninputtoanoutputusingfuzzylogic.Theprocessinvolvesallthepieces:membershipfunctions,fuzzylogicoperators,andif-thenrules.Resultsofexperimentsshowsthatourframeworkismoreefficientthanotherframeworksandapproachesforriskassessmentinsoftwareprojects.AssessmentofEffectiveRiskinSoftwareProjectsbasedonWallace’sClassificationUsing59FuzzyLogicCopyright©2013MECSI.J.InformationEngineeringandElectronicBusiness,2013,4,58-64Therestofthispaperisorganizedasfollows:Section2describestherelatedworks.Section3describestheconceptsanddefinitions.Section4explainstheWallace’sWork.Section5describesproposedapproachandfinallysection6concludesthepaperwithsomediscussion.II.RELATEDWORKThissectionpresentsworksthatrelatedtoriskandriskmanagement.Boehm[2]proposedasoftwareriskmanagementframework.Heidentifiedalistofthetop-tensoftwarerisksbasedonhisexperienceatTRW.Thereweresomeproblemsinhissurvey.Thelistoftop-tensoftwareriskslackedatheoreticalfoundation.Secondly,theserisksaresetaccordingtosoftwaredevelopmentenvironmentin1991butscaleanddiversityofsoftwarehaveincreasedandthus,thelisthasbecomeinadequate.Barkietal.[4]conductedasurveyinQuebec.Theyidentifiedalistwhichincluded23softwarerisks.Theyclassifiedthemintofivegroups.ThelistprovidedacomprehensibleinstrumentbutWallaceetal.[13]explainedthattheassessmentscaleofeachriskwascomplex.Schmidtetal.conductedaDelphisurveytoreducethebiasofasingle-cultureviewpoint.Theyintegratedtheoptions