




版權(quán)說(shuō)明:本文檔由用戶提供并上傳,收益歸屬內(nèi)容提供方,若內(nèi)容存在侵權(quán),請(qǐng)進(jìn)行舉報(bào)或認(rèn)領(lǐng)
文檔簡(jiǎn)介
MultipleRegressionDr.AndyFieldSlide2AimsUnderstandWhenToUseMultipleRegression.Understandthemultipleregressionequationandwhatthebetasrepresent.UnderstandDifferentMethodsofRegressionHierarchicalStepwiseForcedEntryUnderstandHowtodoaMultipleRegressiononPASW/SPSSUnderstandhowtoInterpretmultipleregression.UnderstandtheAssumptionsofMultipleRegressionandhowtotestthemSlide3WhatisMultipleRegression?LinearRegressionisamodeltopredictthevalueofonevariablefromanother.MultipleRegressionisanaturalextensionofthismodel:Weuseittopredictvaluesofanoutcomefromseveralpredictors.Itisahypotheticalmodeloftherelationshipbetweenseveralvariables.Regression:AnExampleArecordcompanybosswasinterestedinpredictingrecordsalesfromadvertising.Data200differentalbumreleasesOutcomevariable:Sales(CDsandDownloads)intheweekafterreleasePredictorvariablesTheamount(in£s)spentpromotingtherecordbeforerelease(seelastlecture)Numberofplaysontheradio(newvariable)Slide5TheModelwithOnePredictorSlide6MultipleRegressionasanEquationWithmultipleregressiontherelationshipisdescribedusingavariationoftheequationofastraightline.Slide7b0b0
istheintercept.TheinterceptisthevalueoftheYvariablewhenallXs=0.ThisisthepointatwhichtheregressionplanecrossestheY-axis(vertical).Slide8BetaValuesb1
istheregressioncoefficientforvariable1.b2
istheregressioncoefficientforvariable2.bn
istheregressioncoefficientfornthvariable.Slide9TheModelwithTwoPredictorsbAdvertsbairplayb0Slide10MethodsofRegressionHierarchical:Experimenterdecidestheorderinwhichvariablesareenteredintothemodel.ForcedEntry:Allpredictorsareenteredsimultaneously.Stepwise:Predictorsareselectedusingtheirsemi-partialcorrelationwiththeoutcome.Slide12HierarchicalRegressionKnownpredictors(basedonpastresearch)areenteredintotheregressionmodelfirst.Newpredictorsarethenenteredinaseparatestep/block.Experimentermakesthedecisions.Slide13HierarchicalRegressionItisthebestmethod:Basedontheorytesting.Youcanseetheuniquepredictiveinfluenceofanewvariableontheoutcomebecauseknownpredictorsareheldconstantinthemodel.BadPoint:Reliesontheexperimenterknowingwhatthey’redoing!Slide14ForcedEntryRegressionAllvariablesareenteredintothemodelsimultaneously.Theresultsobtaineddependonthevariablesenteredintothemodel.Itisimportant,therefore,tohavegoodtheoreticalreasonsforincludingaparticularvariable.Slide15StepwiseRegressionIVariablesareenteredintothemodelbasedonmathematicalcriteria.Computerselectsvariablesinsteps.Step1SPSSlooksforthepredictorthatcanexplainthemostvarianceintheoutcomevariable.ExamPerformanceRevisionTimeVarianceaccountedforbyRevisionTime(33.1%)PreviousExamVarianceexplained(1.7%)DifficultyVarianceexplained(1.3%)Slide18StepwiseRegressionIIStep2:Havingselectedthe1stpredictor,asecondoneischosenfromtheremainingpredictors.Thesemi-partialcorrelationisusedasacriterionforselection.Slide19Semi-PartialCorrelationPartialcorrelation:measurestherelationshipbetweentwovariables,controllingfortheeffectthatathirdvariablehasonthemboth.Asemi-partialcorrelation:Measurestherelationshipbetweentwovariablescontrollingfortheeffectthatathirdvariablehasononlyoneoftheothers.Slide20ExamAnxietyExamAnxietyRevisionRevisionPartial
CorrelationSemi-PartialCorrelationSlide21Semi-PartialCorrelationinRegressionThesemi-partialcorrelationMeasurestherelationshipbetweenapredictorandtheoutcome,controllingfortherelationshipbetweenthatpredictorandanyothersalreadyinthemodel.Itmeasurestheuniquecontributionofapredictortoexplainingthevarianceoftheoutcome.Slide22Slide23ProblemswithStepwiseMethods
Relyonamathematicalcriterion.VariableselectionmaydependupononlyslightdifferencesintheSemi-partialcorrelation.Theseslightnumericaldifferencescanleadtomajortheoreticaldifferences.Shouldbeusedonlyforexploration
Slide24DoingMultipleRegressionSlide25DoingMultipleRegressionRegressionStatisticsRegressionDiagnosticsSlide28Output:ModelSummarySlide29RandR2
RThecorrelationbetweentheobservedvaluesoftheoutcome,andthevaluespredictedbythemodel.R2Yheproportionofvarianceaccountedforbythemodel.Adj.R2AnestimateofR2inthepopulation(shrinkage).Slide30Output:ANOVASlide31AnalysisofVariance:ANOVATheF-testlooksatwhetherthevarianceexplainedbythemodel(SSM)issignificantlygreaterthantheerrorwithinthemodel(SSR).Ittellsuswhetherusingtheregressionmodelissignificantlybetteratpredictingvaluesoftheoutcomethanusingthemean.Slide32Output:betasSlide33HowtoInterpretBetaValues
Betavalues:thechangeintheoutcomeassociatedwithaunitchangeinthepredictor.Standardisedbetavalues:tellusthesamebutexpressedasstandarddeviations.Slide34BetaValuesb1=0.087.So,asadvertisingincreasesby£1,recordsalesincreaseby0.087units.b2=3589.So,eachtime(perweek)asongisplayedonradio1itssalesincreaseby3589units.Slide35ConstructingaModelSlide36StandardisedBetaValues
1=0.523Asadvertisingincreasesby1standarddeviation,recordsalesincreaseby0.523ofastandarddeviation.
2=0.546Whenthenumberofplaysonradioincreasesby1s.d.itssalesincreaseby0.546standarddeviations.Slide37InterpretingStandardisedBetasAsadvertisingincreasesby£485,655,recordsalesincreaseby0.52380,699=42,206.Ifthenumberofplaysonradio1perweekincreasesby12,recordsalesincreaseby0.54680,699=44,062.ReportingtheModelSlide39HowwelldoestheModelfitthedata?
Therearetwowaystoassesstheaccuracyofthemodelinthesample:ResidualStatisticsStandardizedResidualsInfluentialcasesCook’sdistanceSlide40StandardizedResiduals
Inanaveragesample,95%ofstandardizedresidualsshouldliebetween2.99%ofstandardizedresidualsshouldliebetween2.5.OutliersAnycaseforwhichtheabsolutevalueofthestandardizedresidualis3ormore,islikelytobeanoutlier.Slide41Cook’sDistanceMeasurestheinfluenceofasinglecaseonthemodelasawhole.Weisberg(1982):Absolutevaluesgreaterthan1maybecauseforconcern.Slide42Generalization
Whenwerunregression,wehopetobeabletogeneralizethesamplemodeltotheentirepopulation.Todothis,severalassumptionsmustbemet.Violatingtheseassumptionsstopsusgeneralizingconclusionstoourtargetpopulation.Slide43StraightforwardAssumptions
VariableType:OutcomemustbecontinuousPredictorscanbecontinuousordichotomous.Non-ZeroVariance:Predictorsmustnothavezerovariance.Linearity:Therelationshipwemodelis,inreality,linear.Independence:Allvaluesoftheoutcomeshouldcomefromadifferentperson.Slide44TheMoreTrickyAssumptionsNoMulticollinear
溫馨提示
- 1. 本站所有資源如無(wú)特殊說(shuō)明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請(qǐng)下載最新的WinRAR軟件解壓。
- 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請(qǐng)聯(lián)系上傳者。文件的所有權(quán)益歸上傳用戶所有。
- 3. 本站RAR壓縮包中若帶圖紙,網(wǎng)頁(yè)內(nèi)容里面會(huì)有圖紙預(yù)覽,若沒(méi)有圖紙預(yù)覽就沒(méi)有圖紙。
- 4. 未經(jīng)權(quán)益所有人同意不得將文件中的內(nèi)容挪作商業(yè)或盈利用途。
- 5. 人人文庫(kù)網(wǎng)僅提供信息存儲(chǔ)空間,僅對(duì)用戶上傳內(nèi)容的表現(xiàn)方式做保護(hù)處理,對(duì)用戶上傳分享的文檔內(nèi)容本身不做任何修改或編輯,并不能對(duì)任何下載內(nèi)容負(fù)責(zé)。
- 6. 下載文件中如有侵權(quán)或不適當(dāng)內(nèi)容,請(qǐng)與我們聯(lián)系,我們立即糾正。
- 7. 本站不保證下載資源的準(zhǔn)確性、安全性和完整性, 同時(shí)也不承擔(dān)用戶因使用這些下載資源對(duì)自己和他人造成任何形式的傷害或損失。
最新文檔
- 中石化買賣石油合同范本
- 書(shū)刊供貨合同范本
- 廠房 設(shè)備維修合同范本
- 網(wǎng)上調(diào)查課題申報(bào)書(shū)
- 合同范本組成
- 保潔小區(qū)開(kāi)荒合同范本
- 醫(yī)用銷售合同范本
- 員工借調(diào)合同范例
- 產(chǎn)品模特簽約合同范本
- 南寧雅閣購(gòu)車合同范本
- 重慶市屬事業(yè)單位招聘真題2024
- 新版第三類醫(yī)療器械分類目錄
- 多智能體機(jī)器人系統(tǒng)控制及其應(yīng)用課件全套第1-8章多智能體機(jī)器人系統(tǒng)-異構(gòu)多智能體系統(tǒng)的協(xié)同控制和最優(yōu)控制
- 弗洛姆異化理論
- AQL抽樣標(biāo)準(zhǔn)表xls2
- 碳納米管_ppt課件
- 【課件】第2課如何鑒賞美術(shù)作品課件-高中美術(shù)人教版(2019)美術(shù)鑒賞
- 人力資源部經(jīng)理崗位說(shuō)明書(shū)
- [康熙字典9畫(huà)五行屬金的字加解釋] 康熙字典五行屬金的字
- 液化氣罐定期檢驗(yàn)方案
- 關(guān)于老年癡呆癥及其智能陪護(hù)設(shè)備的調(diào)查報(bào)告
評(píng)論
0/150
提交評(píng)論