搜索
2017APMCMsummarysheetThequalityofsleepisimportanttothehealthofthebody.Anincreasingnumberofstudieshaveshownthatpoorsleepmaynegativelyaffectourphysicalandmentalwell-being.Therefore,weneedtokeepagoodsleepstatetomaintainourphysicalhealth.Thefactorsthatinfluenceoursleepqualityarevaried,Thecoreobjectiveofthispaperistoselectthevariablesthatarehighlycorrelatedwiththequalityofsleep,Onthisbasis,theclassificationrelationshipbetweenthevariablesandthediagnosisresultsisstudied,Italsopredictsthepossiblediagnosisofthesamplesinannexiii,Finally,makeareasonablesleepplantoprotectpeople'shealth.Inthefirstquestion,We'llstartbyusingthespearmancorrelationcoefficienttoselectthevariablesthathavesignificantcorrelation,andthentheSPSSsoftwarewasadoptedtoestablishmultiplelogisticmodels,thelastcategoryofsleepqualitywasusedasthereference,andestimatedtheconversionprobabilityofthecorrespondingsleepqualitycategorywhenthechangeofvariablevaluewas1.Andthechangedirectionanddegreeofchangeofsleepqualityareevaluatedbytheprobabilityvalue,Finally,theresultsofcorrelationbetweenvariablesandsleepqualitywereobtained.Inthesecondquestion,boththediagnosisresultsandsleeparemultidimensionalclassificationvariables,Tosolvethisquestion,weusethecorrespondinganalysismethod,whicheffectivelycombininglistanalysiswithfactoranalysis.ThecorrelationbetweendiagnosisandsleepinvariousstateswasfoundbyRowandColumnpoints.Inthethirdquestion,Weusedthedecision-treemodeltoexplorethelinkbetweendiagnosisresultsandage,gender,andsleepindicatorstodeterminethedegreeofassociationbetweeneachindicatorandthediagnosis,andtherulesetwasestablishedby45percent.Then,accordingtotheruleset,thedatainannexiiiisdiagnosed,andthecorrectdiagnosiswasgreaterthan45%.Anditwasfoundthatwhenthesamplesizecorrespondingtovariousdiagnosticresultsincreasedcorrespondingly,thepredictedaccuracyincreased.Inthefourthquestion,basedontheliteratureandthepracticalfactors,wemakeaplanwhichisaboutscientificsleeptimeandsleeplengthforpeoplewhohavebadsleepqualityorbadsleepefficiency.Byimprovingthesleepdurationandlengthofpatientsinannexii,thedataofdiagnosisresultswereobservedtoverifytheeffectivenessoftheplan.Theresultsshowedthatpeoplewhosleptwellwereabletofallasleepat10:40andspentsixtoeighthoursinsleeping.Peoplewithpoorsleepqualityandefficiencyneedtogetintosleepby10o'clockandsleepforeighthoursKeywords:multiplelogisticmodelsdecision-treemodelcorrespondinganalysisTeam#0001Team#0001Page1of21.ProblemDescription1.1BackgroundThequalityofsleepcanhaveagreatimpactonourphysicalhealthanddailylife.Usually,aday'smentalstatedependsonthequalityofthenightbefore.Ahighqualitysleepwillensurethatthenextdayisfullofenergy.However,theincidenceofinsomniainouradultpopulationisstillhigh,ashighas38.2%,andtheincreasingsleepproblemhasarousedwidespreadconcern.Ascientificsleeplifeisessentialtoagoodlife.1.2TheproblemtobesolvedProblem1:Identifyindicatorsthataffectsleepqualityandanalyzetheirrelevance.Problem2:Therelationshipbetweendiagnosisandsleepwasdiscussed,includingsevenindicators:sleepquality,sleepduration,lengthofsleep,andsleepefficiency.Problem3:Accordingtotherelationshipbetweendiagnosisandsleep,sexandage,thedatainAnnexIIIwerepredicted,andthediagnosisresultswereobtainedbyerror.Problem4:Accordingtotheaboveanalysis,ascientificsleepplanisdevelopedandresultsofitseffectivenessaretested.2.ProblemAnalysis2.1Analysisofproblem1TheproblemrequiresthatweanalyzethevariablesassociatedwiththequalityofsleepinAnnexI.ItcanbeseenfromthedataofAnnexI,sleepqualityisacategoricalvariable,therearefourtypesofvalues:0,1,2,and3,ofthese,0representsthebestsleepand3theworst.Optionalvariablesareage,sex,source,sleepquality,reliability,psychoticism,nervousness,character.Amongthem,inadditiontosexandage,itisacontinuousvariable.Therefore,theproblemistransformedintothecorrelationbetweenanorderedmulti-classificationvariableasoutput,continuousvariablesandcategoricalvariablesasindependentvariables.Tosolvetheproblem,thespearmancoefficientorderedlogisticregressionmodelwasadopted.2.2Analysisofproblem2Problem2requiresustostudytherelationshipbetweendiagnosisresultsandsleepaccordingtoAnnexII.InAnnexII,sevenordervariablesaregivenforsleep,0isthebest,and3istheworst,thisisananalysisoftwogroupsoforderedclassvariables.Inordertosolvethisproblem,weneedtochooseasuitablefororderlylevelvariablerelationshipminingmodel.Consideringthediagnosisresulttypeandsleepindexnumberismany,andthediagnosiswithindicatorsarelikelytoexistbetweenthesevensleepdependencies,sothispapertakesthecorrespondenceanalysismodel.Firstly,theoriginaldatastructurewasanalyzedbymeansofhealthdimension,andatthesametime,analysisthediagnosisresultsandtherelationshipbetweenthesevenindicatorsofsleep.2.3Analysisofproblem3Team#0001Page2of2Problem3asksustopredictthediagnosisofseveralpeopleinAnnexIII.Inordertoaccuratelypredictthediagnosis,weneedtostudythediagnosisandsleepfactors,theaccuracyoftherelationshipbetweenage,sex,andundertheconditionofthepermittederror,accordingtothemodeltosimulatetheAnnexIIIdiagnosisprediction.Problem3takesthedecisiontreemodel,whichcan