Konsepti
Regressioanalyysi onmatemaattinen malli. Kun riippuvalla muuttujalla ja riippumattomalla muuttujalla on lineaarinen suhde, se onerityinen lineaarinen malli.
Yksinkertaisin tapaus on yksimuuttujalineaarinen regressio, joka koostuu riippumattomasta muuttujasta ja riippuvaisesta muuttujasta, jotka ovat suurin piirtein lineaarisesti sukua; malliY=a+bX+ε(Xisriippumaton muuttuja,Yissyymuuttuja,εon satunnainen virhe).
Usuallyassumethatthemeanvalueoftherandomerroris0,andthevarianceisσ^2(σ^2﹥0,σ^2hasnothingtodowiththevalueofX).Ifitisfurtherassumedthattherandomerrorfollowsanormaldistribution,itiscalledanormallinearmodel.Generally,iftherearekindependentvariablesand1dependentvariable,thevalueofthedependentvariableisdividedintotwoparts:onepartisaffectedbytheindependentvariable,thatis,expressedasitsfunction,thefunctionformisknownandcontainsunknownparameters;theotherpartisdeterminedbyOtherunconsideredfactorsandrandomeffectsarerandomerrors.
Whenthefunctionisalinearfunctionwithunknownparameters,itiscalledalinearregressionanalysismodel;whenthefunctionisanonlinearfunctionwithunknownparameters,itiscalledanonlinearregressionanalysismodel.Whenthenumberofindependentvariablesisgreaterthan1,itiscalledmultipleregression,andwhenthenumberofdependentvariablesisgreaterthan1,itiscalledmultipleregression.
Regressioanalyysin sisältö
Regressioanalyysin pääsisältö on seuraava:
①Startingfromasetofdata,determinethequantitativerelationshipbetweencertainvariables;Thatis,amathematicalmodelisestablishedandunknownparametersareestimated.Usuallytheleastsquaremethodisused.
②Testaasuhteiden luotettavuus.
③Intherelationshipbetweenmultipleindependentvariablesaffectingadependentvariable,judgewhethertheindependentvariablehasasignificantimpact,andselectthesignificantimpactintothemodel,andeliminateinsignificantvariables.Stepwiseregression,forwardregression,andbackwardregressionareusuallyused.
④Usetherequiredrelationshiptopredictorcontrolacertainprocess.
Theapplicationofregressionanalysisisveryextensive,andtheuseofstatisticalsoftwarepackagescanmakevariousalgorithmsmoreconvenient.
Regression tyypit
Regression päätyypit ovat:lineaarinen regressio, käyräviivainen regressio, binääriloginen regressio ja moniloginen regressio.
Applicationofanalysis
Correlationanalysisstudiesthecorrelationbetweenphenomena,thedirectionandclosenessofcorrelation,andgenerallydoesnotdistinguishbetweenindependentvariablesordependentvariables.Regressionanalysisistoanalyzethespecificformsofcorrelationbetweenphenomena,determinethecausalrelationship,andusemathematicalmodelstoexpressthespecificrelationship.Forexample,fromthecorrelationanalysis,wecanknowthatthe"quality"and"usersatisfaction"variablesarecloselyrelated,butwhichvariablebetweenthesetwovariablesisaffectedbywhichvariable,andthedegreeofinfluence,requiresregressionanalysisMethodtodetermine.
Generallyspeaking,regressionanalysisistodeterminethecausalrelationshipbetweendependentvariablesandindependentvariables,establisharegressionmodel,andsolvethevariousparametersofthemodelbasedonthemeasureddata,andthenevaluatewhethertheregressionmodelisItcanfitthemeasureddatawell;ifitcanfitwell,furtherpredictionscanbemadebasedontheindependentvariables.
Forexample,ifyouwanttostudythecausalrelationshipbetweenqualityandusersatisfaction,inapracticalsense,productqualitywillaffectusersatisfaction,sosetusersatisfactionasthedependentvariableandrecorditasY;Qualityistheindependentvariable,denotedasX.AccordingtothescatterplotinFigure8-3,thefollowinglinearrelationshipcanbeestablished:
Y=A+BX+§
where:AandBareundeterminedparameters,andAisregressionTheinterceptofthestraightline;Bistheslopeoftheregressionline,whichrepresentstheaveragechangeofYwhenXchangesbyoneunit;§istherandomerrortermthatdependsonusersatisfaction.
LinearregressioncanbeeasilyimplementedintheSPSSsoftware.Theregressionequationisasfollows:
y = 0,857+0,836xTheintercepTegressionLineOnTheyAxisis0.857AndTheSlopeis0.836, se, forveryOnOnPointInProvementInquality, käyttäjättyypintainen
Theexampleshownaboveisasimplelinearregressionproblemofoneindependentvariable.Duringdataanalysis,thiscanalsobeextendedtomultipleregressionofmultipleindependentvariables.PleaserefertothespecificregressionprocessandmeaningRefertorelevantstatisticsbooks.Inaddition,intheSPSSresultoutput,R2,FtestvalueandTtestvaluecanalsobereported.R2isalsocalledthecoefficientofdeterminationoftheequation,whichindicatesthedegreeofinterpretationofthevariableXtoYintheequation.ThevalueofR2isbetween0and1.Thecloserto1,thestrongertheinterpretationabilityofXtoYintheequation.R2isusuallymultipliedby100%toexpressthepercentageofYchangeexplainedbytheregressionequation.TheFtestisoutputthroughtheanalysisofvariancetable,andthesignificancelevelisusedtotestwhetherthelinearrelationshipoftheregressionequationissignificant.Generallyspeaking,significancelevelsbelow0.05aremeaningful.WhentheFtestpasses,itmeansthatatleastoneoftheregressioncoefficientsintheequationissignificant,butnotallregressioncoefficientsaresignificant,soaTtestisneededtoverifythesignificanceoftheregressioncoefficients.Similarly,theTtestcanbedeterminedbythesignificanceleveloralook-uptable.Intheexampleshownabove,themeaningofeachparameterisshowninTable1-1.
Taulukko 1-1linearregressionequationtest
indeksi | > Arvo | Näkyvyystaso | > Merkitys |
> R | > 0,89 | "Laatu" selittää89 %"Käyttäjätyytyväisyyden" muutoksen astetta | |
> F | > 276,82 | > 0,001 | Regressioyhtälön lineaarinen suhde onmerkittävä |
> T | > 16.64 | > 0,001 | Regressioyhtälön kerroin onmerkittävä |