AJavatoolforminingMoodledata
R.PEDRAZA-PEREZ,
C.ROMEROANDS.VENTURAUniversityofCordoba,Spain ________________________________________________________________________ ThispapershowsaJavaMoodleminingtoolorientednotonlyforexpertsindataminingbutalsoforerslikeinstructorsandcoursewareauthors.Thistoolhasawizardwithasimpleinterfacebasedonbigfriendlybuttonsanddefaultoptionstofacilitatetheexecutionofallthestepsinthedataminingprocess.KeyWordsandPhrases:EducationalDataMining,Moodle,DataMiningTool,Wizard. ________________________________________________________________________
1.INTRODUCTIONANDDESCRIPTIONOFTHEMININGTOOLGeneraldataminingtoolstodayrangefrommercial,suchasDBMiner,SPSSClementine,SASEnterpriseMiner,IBMIntelligentMiner,toopensources,likeWekaandRapidMiner.However,allthesetoolsarenotspecificallydesignedforpedagogical/educationalpurposessotheyarecumbersomeforaneducatortousebecausetheyarenormallydesignedmoreforpowerandflexibilitythanforsimplicity(RomeroandVentura,2010).Inordertoresolvethisproblem,wehavedevelopedaJavadata-miningtoolspecificallyforminingMoodledata.WehavechosenMoodlebecauseitisoneofthemostfrequentlyusedLearningManagementSystems(LMS).ThistoolhasbeenimplementedinJavausingtheKEELframework(Alcalaetal.,2009)whichisanopensourceframeworkforbuildingdataminingmodels.OurMoodleminingtoolprovidesawizardthathasasimpleinterfacebasedonbigfriendlybuttons(seeFigure1)tofacilitateitsexecution. Fig.1.MainwindowoftheMoodleminingtoolinterface. AsFigure1shows,ourwizardguidesusersthroughthemonstepsinvolvedinanyknowledgediscoveryprocess,whicharedescribedbelow. Tocreatedatafiles.Firstofall,usersmustcreatedatafilesstartingfromaMoodledatabaseorlogfilesprovidedbyMoodle.Then,theycanchoosefromamongdifferentoptions:creatingadatafilefromoneparticularMoodletable,creatingasummarydatafilefromdifferentMoodletables,orcreatingadatafile __________________________________________________________________________________________
Authors’addresses:
R.Pedraza-Perez,
C.Romero,
S.Ventura;DepartmentofComputerScience,UniversityofCordoba,Spain;E-mail:i22peper@uco.es,cromero@uco.es,sventura@uco.es; fromoneMoodlelogfile.Thesummarizationoptionallowsfortheintegrationofallthemostimportantstudentinformationduringafullcourse,sincestudentandinteractiondataarespreadoverseveralMoodletables.Attheendofthisprocess,atextdatafileissaved/createdusingtheKEELdataformat(Alcalaetal.,2009).Topre-processdatafiles.ThismoduleallowsuserstomodifyKEELdatafiles.Theapplicationprovidesseveralpre-processingtaskssuchas:discretization(todividenumericalvaluesintocategoricalvalues),edition(toaddordeletecolumnsorrows,forexampletoaddstudents’finalmarks)anddivision(tosplitdatafileintotrainingandtestdatafiles).Tovisualizedatafiles.Thismodulewillshow(itisnowunderconstruction)differenttypesofgraphicalrepresentationsoftheaforementioneddata.Tominedatafiles.Finally,userscanapplydataminingtechniques.Ourminingtoolprovidesalargenumberofalgorithmsgroupedintothreemethods/tasks:classification,regression,andassociation.Firstly,usershavetoselectthedesireddataminingmethodandthenoneoftheavailablealgorithms.Ourwizardalwaysmendsonealgorithmanditsparametersbydefaultforeachmethodinordertofacilitateitsusage/executionforbeginners.Userscanchangethecurrentselectiontoanyotheralgorithmandchangeitsparameters.Figure2showstheexecutionoftheC4.5algorithmoverasummarydatafileand10-foldcross-validation.Anyresultingfiles(.traand.testfilesthatcontainpartialclassificationresults,andthe.txtfilecontainingthemodelobtained)canbeselectedandvisualizedeasily(seeFigure2ontherighthandside).Inthisexample,themodelobtainedisadecisiontreethatusesIF-THENrulesandalsoshowsasummarywiththenumberofnodesandleavesonthetree,thenumberandpercentageofcorrectlyandincorrectlyclassifiedinstances,etc. Fig.2.Somewindowsinthedataminingwizard. ACKNOWLEDGMENTSThisresearchissubsidizedbyprojectsoftheRegionalGovernmentofAndaluciaandtheMinistryofScienceandTechnology,P08-TIC-3720andTIN2008-06681-C06-03respectively,andFEDERfunds. REFERENCES ALCALÁ,
J.,SÁNCHEZ,
L.,GARCÍ
A,S.,DELJESUS,
M.ET.AL.2009.KEELasoftwaretooltoassessevolutionaryalgorithmstodataminingproblems.SoftComputing,13
(3),307-318. ROMEROC,VENTURA,
S.EducationalDataMining:AReviewoftheState-of-the-Art.2010.IEEETransactionsonSystems,Manandics,partC:ApplicationsandReviews,40
(6),601-618.