Regression - Binary Logit - Q Wiki

The Binary Logit is a form of regression analysis that models a binary dependent variable (e.g. yes/no, pass/fail, win/lose). Regression-BinaryLogit FromQ Jumptonavigation Jumptosearch Modelabinarydependentvariable(e.g.,yes/no,pass/fail,win/lose).AlsoknownasaLogisticregressionorBinomialregression. TheBinaryLogit[1]isaformofregressionanalysisthatmodelsabinarydependentvariable(e.g.yes/no,pass/fail,win/lose).ItisalsoknownasaLogisticregression,andBinomialregression. Dataformat Thekeyrequirementforabinarylogitregressionisthatthedependentvariableisbinary.InDisplayr,thebestdataformatforthistypeis“Nominal:Mutuallyexclusivecategories”,withvaluesof“0”and“1”. Theindependentvariablescanbecontinuous,categorical,orbinary—justaswithanyotherregressionmodel. Interpretation Variablestatisticsmeasuretheimpactandsignificanceofindividualvariableswithinamodel,whileoverallstatisticsapplytothemodelasawhole.Bothareshowninthebinarylogitoutput. Variablestatistics Estimatethemagnitudeofthecoefficientindicatesthesizeofthechangeintheindependentvariableasthevalueofthedependentvariablechanges.Apositivenumberindicatesadirectrelationship(yincreasesasxincreases),andanegativenumberindicatesaninverserelationship(ydecreasesasxincreases. Thecoefficientiscoloredifthevariableisstatisticallysignificantatthe5%level. StandardErrormeasurestheaccuracyofanestimate.Thesmallerthestandarderror,themoreaccuratethepredictions. Z-valuetheestimatedividedbythestandarderror.Themagnitude(eitherpositiveornegative)indicatesthesignificanceofthevariable.Thevaluesarehighlightedbasedontheirmagnitude. P-valueexpressesthez-valueasaprobability.Ap-valueunder0.05meansthatthevariableisstatisticallysignificantatthe5%level;ap-valueunder0.01meansthatthevariableisstatisticallysignificantatthe1%level.P-valuesunder0.05areshowninbold. Overallstatistics nthesamplesizeofthemodel McFadden’srho-squaredassessthegoodnessoffitofthemodel.Alargernumberindicatesthatthemodelcapturesmoreofthevariationinthedependentvariable. AICAkaikeinformationcriterionisameasureofthequalityofthemodel.Whencomparingsimilarmodels,theAICcanbeusedtoidentifythesuperiormodel. SeealsoRegressionDiagnostics. Example Theexamplebelowisamodelthatpredictsasurveyrespondent’slikelihoodofhavingconsumedafast-foodproductbasedoncharacteristicslikeage,gender,andworkstatus. CreateaBinaryLogitModelinDisplayr 1.GotoInsert>Regression>BinaryLogit 2.UnderInputs>Outcome,selectyourdependentvariable 3.UnderInputs>Predictor(s),selectyourindependentvariables ObjectInspectorOptions OutcomeThevariabletobepredictedbythepredictorvariables. PredictorsThevariable(s)topredicttheoutcome. AlgorithmThefittingalgorithm.DefaultstoRegressionbutmaybechangedtoothermachinelearningmethods. Type:Youcanusethisoptiontotogglebetweendifferenttypesofregressionmodels,butnotethatcertaintypesarenotappropriateforcertaintypesofoutcomevariable. LinearAppropriateforacontinuousoutcomevariable.SeeRegression-LinearRegression. BinaryLogitAppropriateiftheoutcomeisbinary(i.e.fallsinoneoftwocategories).SeeRegression-BinaryLogit. OrderedLogitAppropriateforadiscreteoutcomewherethecategorieshaveanaturalorder(e.g.Low,Medium,High).SeeRegression-OrderedLogit. MultinomialLogitAppropriateforadiscreteoutcomewithunorderedcategories.SeeRegression-MultinomialLogit. PoissonAppropriateforcountoutcomes(i.e.outcomesthattakeonlypositiveintegervalues).SeeRegression-PoissonRegression. Quasi-PoissonAppropriateforcountoutcomes.SeeRegression-Quasi-PoissonRegression. NBDAppropriateforcountoutcomes.SeeRegression-NBDRegression. RobuststandarderrorsComputesstandarderrorsthatarerobusttoviolationsoftheassumptionofconstantvariance(i.e.,heteroscedasticity).SeeRobustStandardErrors.ThisisonlyavailablewhenTypeisLinear. MissingdataSeeMissingDataOptions. Output SummaryThedefault;asshownintheexampleabove. DetailTypicalRoutput,someadditionalinformationcomparedtoSummary,butwithouttheprettyformatting. ANOVAAnalysisofvariancetablecontainingtheresultsofChi-squaredlikelihoodratiotestsforeachpredictor. RelativeImportanceAnalysisTheresultsofarelativeimportanceanalysis.Seehereandthereferencesformoreinformation.ThisoptionisnotavailableforMultinomialLogit.Notethatcategoricalpredictorsarenotconvertedtobenumeric,unlikeinDriver(Importance)Analysis-RelativeImportanceAnalysis. ShapleyRegressionSeehereandthereferencesformoreinformation.ThisoptionisonlyavailableforLinearRegression.Notethatcategoricalpredictorsarenotconvertedtobenumeric,unlikeinDriver(Importance)Analysis-Shapley. JaccardCoefficientComputestherelativeimportanceofthepredictorvariablesagainsttheoutcomevariablewiththeJaccardCoefficients.SeeDriver(Importance_Analysis-JaccardCoefficient.Thisoptionrequiresbothbinaryvariablesfortheoutcomevariableandthepredictorvariables. CorrelationComputestherelativeimportanceofthepredictorvariablesagainsttheoutcomevariableviathebivariatePearsonproductmomentcorrelations.SeeDriver(Importance)Analysis-Correlationandreferencesthereinformoreinformation. EffectsPlotPlotstherelationshipbetweeneachofthePredictorsandtheOutcome.NotavailableforMultinomialLogit. CorrectionThemultiplecomparisonscorrectionappliedwhencomputingthep-valuesofthepost-hoccomparisons. VariablenamesDisplaysVariableNamesintheoutputinsteadoflabels. AbsoluteimportancescoresWhethertheabsolutevalueofRelativeImportanceAnalysisscoresshouldbedisplayed. AuxiliaryvariablesVariablestobeusedwhenimputingmissingvalues(inadditiontoalltheothervariablesinthemodel). Weight.WhereaweighthasbeensetfortheROutput,itwillautomaticallyappliedwhenthemodelisestimated.Bydefault,theweightisassumedtobeasamplingweight,andthestandarderrorsareestimatedusingTaylorserieslinearization(bycontrast,intheLegacyRegression,weightcalibrationisused).SeeWeights,EffectiveSampleSizeandDesignEffects. FilterThedataisautomaticallyfilteredusinganyfilterspriortoestimatingthemodel. CrosstabInteractionOptionalvariabletotestforinteractionwithothervariablesinthemodel.Theinteractionvariableistreatedasacategoricalvariable.Coefficientsinthetablearecomputedbycreatingseparateregressionsforeachleveloftheinteractionvariable.Toevaluatewhetheracoefficientissignificantlyhigher(blue)orlower(red),weperformat-testofthecoefficientcomparedtothecoefficientusingtheremainingdataasdescribedinDriverAnalysis.P-valuesarecorrectedformultiplecomparisonsacrossthewholetable(excludingtheNETcolumn).TheP-valueinthesub-titleiscalculatedusingathelikelihoodratiotestbetweenthepooledmodelwithnointeractionvariable,andamodelwhereallpredictorsinteractwiththeinteractionvariable. AutomatedoutlierremovalpercentageAnumericvaluebetween0and50(including0butnot50)isusedtospecifythepercentageofthedatathatisremovedfromanalysisduetooutliers.AllregressiontypesexceptforthecaseofMultinomialLogitsupportthisfeature.Ifazero-valueisselectedforthisinputcontrolthennooutlierremovalisperformedandastandardregressionoutputfortheentire(possiblyfiltered)datasetisapplied.Ifanon-zerovalueisselectedforthisoptionthentheregressionmodelisfittedtwice.Thefirstregressionmodelusestheentiredataset(afterfiltershavebeenapplied)andidentifiestheobservationsthatgeneratethelargestresiduals.Theuserspecifiedpercentofcasesinthedatathathavethelargestresidualsarethenremoved.Theregressionmodelisrefittedonthisreduceddatasetandoutputreturned.ThespecificresidualusedvariesdependingontheregressionType. Linear:ThestudentizedresidualinanunweightedregressionandthePearsonresidualinaweightedregression.ThePearsonresidualintheweightedcaseadjustsappropriatelyfortheprovidedsurveyweights. BinaryLogitandOrderedLogit:AtypeofsurrogateresidualfromthesureRpackage(seeGreenwell,McCarthy,BoehmkeandLiu(2018)[2]formoredetails).InBinaryLogititusestheresidsfunctionwiththejitterparametrization.InOrderedLogititusestheresidsfunctionwiththelatentparametrizationtoexploittheorderedlogitstructure. NBDRegression,PoissonRegression:AstudentizeddevianceresidualinanunweightedregressionandthePearsonresidualinaweightedregression. Quasi-PoissonRegression:Atypeofquasi-devianceresidualviatherstudentfunctioninanunweightedregressionandthePearsonresidualinaweightedregression. Thestudentizedresidualcomputesthedistancebetweentheobservedandfittedvalueforeachpointandstandardizes(adjusts)basedontheinfluenceandanexternallyadjustedvariancecalculation.Thestudentizeddevianceresidualcomputesthecontributionthefittedpointhastothelikelihoodandstandardizes(adjusts)basedontheinfluenceofthepointandanexternallyadjustedvariancecalculation(seerstudentfunctioninRandDavisonandSnell(1991)[3]formoredetails).ThePearsonresidualintheweightedcasecomputesthedistancebetweentheobservedandfittedvalueandadjustsappropriatelyfortheprovidedsurveyweights.SeerstudentfunctioninRandDavisonandSnell(1991)formoredetailsofthespecificsofthecalculations. StackdataWhethertheinputdatashouldbestackedbeforeanalysis.Stackingcanbedesirablewheneachindividualinthedatasethasmultiplecasesandanaggregatemodelisdesired.MoreinformationisavailableatStackingDataFiles.IfthisoptionischosenthentheOutcomeneedstobeasingleQuestionthathasaMultitypestructuresuitableforregressionsuchasaPickOne-Multi,PickAnyorNumber-MultiVariableSetthathasaMultitypestructuresuitableforregressionsuchasaBinary-Multi,Nominal-Multi,Ordinal-MultiorNumeric-Multi.Similarly,thePredictor(s)needtobeasingleQuestionthathasaGridtypestructuresuchasaPickAny-GridoraNumber-GridVariableSetthathasaGridtypestructuresuchasaBinary-GridoraNumeric-Grid.Intheprocessofstacking,thedatareductionisinspected.AnyconstructedNETsareremovedunlesscomprisedofsourcevaluesthataremutuallyexclusivetoothercodes,suchastheresultofmergingtwocategories. RandomseedSeedusedtoinitializethe(pseudo)randomnumbergeneratorforthemodelfittingalgorithm.Differentseedsmayleadtoslightlydifferentanswers,butshouldnormallynotmakealargedifference. IncreaseallowedoutputsizeCheckthisboxifyouencounterawarningmessage"TheRoutputhadsizeXXXMB,exceedingthe128MBlimit..."andyouneedtoreferencetheoutputelsewhereinyourdocument;e.g.,tosavepredictedvaluestoaDataSetorexaminediagnostics. Maximumallowedsizeforoutput(MB).ThiscontrolonlyappearsifIncreaseallowedoutputsizeischecked.UseittosetthemaximumallowedsizefortheregressionoutputinMegabytes.ThewarningreferredtoaboveabouttheRoutputsizewillstatetheminimumsizeyouneedtoincreasetotoreturnthefulloutput.Notethathavingverymanylargeoutputsinonedocumentorpagemayslowdowntheperformanceofyourdocumentandincreaseloadtimes. Additionaloptionsareavailablebyeditingthecode. DIAGNOSTICS Plot-Cook'sDistanceCreatesaline/rugplotshowingCook'sDistanceforeachobservation. Plot-Cook'sDistancevsLeverageCreatesascatterplotshowingCook'sdistancevsleverageforeachobservation. Plot-InfluenceIndexCreatesindexplotsofstudentizedresiduals,hatvalues,andCook'sdistance. MulticollinearityTable(VIF)Createsatablecontainingvarianceinflationfactors(VIF)todiagnosemulticollinearity. Plot-NormalQ-QCreatesanormalQuantile-Quantile(QQ)plottorevealdeparturesoftheresidualsfromnormality. Prediction-AccuracyTableCreatesatableshowingtheobservedandpredictedvalues,asaheatmap. TestResidualHeteroscedasticityConductsaheteroscedasticitytestontheresiduals. TestResidualNormality(Shapiro-Wilk)ConductsaShapiro-Wilktestofnormalityonthe(deviance)residuals. Plot-ResidualsvsFittedCreatesascatterplotofresidualsversusfittedvalues. Plot-ResidualsvsLeverageCreatesaplotofresidualsversusleveragevalues. Plot-Scale-LocationCreatesaplotofthesquarerootoftheabsolutestandardizedresidualsbyfittedvalues. TestResidualSerialCorrelation(Durbin-Watson)ConductsaDurbin-Watsontestofserialcorrelation(auto-correlation)ontheresiduals. SAVEVARIABLE(S) FittedValuesCreatesanewvariablecontainingfittedvaluesforeachcaseinthedata. PredictedValuesCreatesanewvariablecontainingpredictedvaluesforeachcaseinthedata. ResidualsCreatesanewvariablecontainingresidualvaluesforeachcaseinthedata. Moreinformation WhatisLogisticRegression? HowtodoLogisticRegressioninDisplayr HowtoInterpretLogisticRegressionOutputs HowtoInterpretLogisticRegressionCoefficients Acknowledgements UsestheglmfromthestatsRpackage.Ifweightsaresupplied,thesvyglmfunctionfromthesurveyRpackageisused.AlsousestheresidsfunctioninfromthesureRpackage.SeealsoRegression-GeneralizedLinearModel. References ↑Yap,J.(2018,August22).Whatislogisticregression?[Blogpost].Accessedfromhttps://www.displayr.com/what-is-logistic-regression/ ↑Greenwell,B.,M.,McCarthy,A.,J.,Boehmke,B.,C.,&Liu,D.(2018).ResidualsandDiagnosticsforBinaryandOrdinalRegressionModels:AnIntroductiontothesurePackage.TheRJournal,10(1),381.https://doi.org/10.32614/rj-2018-004 ↑Davison,A.C.andSnell,E.J.(1991)Residualsanddiagnostics.In:StatisticalTheoryandModelling.InHonourofSirDavidCox,FRS,eds.Hinkley,D.V.,Reid,N.andSnell,E.J.,Chapman&Hall. Code varcontrols=[]; //ALGORITHM varalgorithm=form.comboBox({label:"Algorithm", alternatives:["CART","DeepLearning","GradientBoosting","LinearDiscriminantAnalysis", "RandomForest","Regression","SupportVectorMachine"], name:"formAlgorithm",default_value:"Regression", prompt:"Machinelearningorregressionalgorithmforfittingthemodel"}); controls.push(algorithm); algorithm=algorithm.getValue(); varregressionType=""; if(algorithm=="Regression") { regressionTypeControl=form.comboBox({label:"Regressiontype", alternatives:["Linear","BinaryLogit","OrderedLogit","MultinomialLogit","Poisson", "Quasi-Poisson","NBD"], name:"formRegressionType",default_value:"BinaryLogit", prompt:"Selecttypeaccordingtooutcomevariabletype"}); regressionType=regressionTypeControl.getValue(); controls.push(regressionTypeControl); } //DEFAULTCONTROLS missing_data_options=["Errorifmissingdata","Excludecaseswithmissingdata","Imputation(replacemissingvalueswithestimates)"]; //AMENDDEFAULTCONTROLSPERALGORITHM if(algorithm=="SupportVectorMachine") output_options=["Accuracy","Prediction-AccuracyTable","Detail"]; if(algorithm=="GradientBoosting") output_options=["Accuracy","Importance","Prediction-AccuracyTable","Detail"]; if(algorithm=="RandomForest") output_options=["Importance","Prediction-AccuracyTable","Detail"]; if(algorithm=="DeepLearning") output_options=["Accuracy","Prediction-AccuracyTable","CrossValidation","NetworkLayers"]; if(algorithm=="LinearDiscriminantAnalysis") output_options=["Means","Detail","Prediction-AccuracyTable","Scatterplot","Moonplot"]; if(algorithm=="CART"){ output_options=["Sankey","Tree","Text","Prediction-AccuracyTable","CrossValidation"]; missing_data_options=["Errorifmissingdata","Excludecaseswithmissingdata", "Usepartialdata","Imputation(replacemissingvalueswithestimates)"] } if(algorithm=="Regression"){ if(regressionType=="MultinomialLogit") output_options=["Summary","Detail","ANOVA"]; elseif(regressionType=="Linear") output_options=["Summary","Detail","ANOVA","RelativeImportanceAnalysis","ShapleyRegression","JaccardCoefficient","Correlation","EffectsPlot"]; else output_options=["Summary","Detail","ANOVA","RelativeImportanceAnalysis","EffectsPlot"]; } //COMMONCONTROLSFORALLALGORITHMS varoutputControl=form.comboBox({label:"Output",prompt:"Thetypeofoutputusedtoshowtheresults", alternatives:output_options,name:"formOutput", default_value:output_options[0]}); controls.push(outputControl); varoutput=outputControl.getValue(); if(algorithm=="Regression"){ if(regressionType=="Linear"){ if(output=="JaccardCoefficient"||output=="Correlation") missing_data_options=["Errorifmissingdata","Excludecaseswithmissingdata","Usepartialdata(pairwisecorrelations)"]; else missing_data_options=["Errorifmissingdata","Excludecaseswithmissingdata","Dummyvariableadjustment","Usepartialdata(pairwisecorrelations)","Multipleimputation"]; } else missing_data_options=["Errorifmissingdata","Excludecaseswithmissingdata","Dummyvariableadjustment","Multipleimputation"]; } varmissingControl=form.comboBox({label:"Missingdata", alternatives:missing_data_options,name:"formMissing",default_value:"Excludecaseswithmissingdata", prompt:"Optionsforhandlingcaseswithmissingdata"}); varmissing=missingControl.getValue(); controls.push(missingControl); controls.push(form.checkBox({label:"Variablenames",name:"formNames",default_value:false,prompt:"Displaynamesinsteadoflabels"})); //CONTROLSFORSPECIFICALGORITHMS if(algorithm=="SupportVectorMachine") controls.push(form.textBox({label:"Cost",name:"formCost",default_value:1,type:"number", prompt:"Highcostproducesacomplexmodelwithriskofoverfitting,lowcostproducesasimplermodewithriskofunderfitting"})); if(algorithm=="GradientBoosting"){ controls.push(form.comboBox({label:"Booster", alternatives:["gbtree","gblinear"],name:"formBooster",default_value:"gbtree", prompt:"Boosttreeorlinearunderlyingmodels"})); controls.push(form.checkBox({label:"Gridsearch",name:"formSearch",default_value:false, prompt:"Searchforoptimalhyperparameters"})); } if(algorithm=="RandomForest") if(output=="Importance") controls.push(form.checkBox({label:"Sortbyimportance",name:"formImportance",default_value:true})); if(algorithm=="DeepLearning"){ controls.push(form.numericUpDown({name:"formEpochs",label:"Maximumepochs",default_value:10,minimum:1,maximum:Number.MAX_SAFE_INTEGER, prompt:"Numberofroundsoftraining"})); controls.push(form.textBox({name:"formHiddenLayers",label:"Hiddenlayers",prompt:"Commadelimitedlistofthenumberofnodesineachhiddenlayer",required:true})); controls.push(form.checkBox({label:"Normalizepredictors",name:"formNormalize",default_value:true, prompt:"Normalizetozeromeanandunitvariance"})); } if(algorithm=="LinearDiscriminantAnalysis"){ if(output=="Scatterplot") { controls.push(form.colorPicker({label:"Outcomecolor",name:"formOutColor",default_value:"#5B9BD5"})); controls.push(form.colorPicker({label:"Predictorscolor",name:"formPredColor",default_value:"#ED7D31"})); } controls.push(form.comboBox({label:"Prior",alternatives:["Equal","Observed",],name:"formPrior",default_value:"Observed", prompt:"Probabilitiesofgroupmembership"})); } if(algorithm=="CART"){ controls.push(form.comboBox({label:"Pruning",alternatives:["Minimumerror","Smallesttree","None"], name:"formPruning",default_value:"Minimumerror", prompt:"Removenodesaftertreehasbeenbuilt"})); controls.push(form.checkBox({label:"Earlystopping",name:"formStopping",default_value:false, prompt:"Stopbuildingtreewhenfitdoesnotimprove"})); controls.push(form.comboBox({label:"Predictorcategorylabels",alternatives:["Fulllabels","Abbreviatedlabels","Letters"], name:"formPredictorCategoryLabels",default_value:"Abbreviatedlabels", prompt:"Labellingofpredictorcategoriesinthetree"})); controls.push(form.comboBox({label:"Outcomecategorylabels",alternatives:["Fulllabels","Abbreviatedlabels","Letters"], name:"formOutcomeCategoryLabels",default_value:"Fulllabels", prompt:"Labellingofoutcomecategoriesinthetree"})); controls.push(form.checkBox({label:"Allowlong-runningcalculations",name:"formLongRunningCalculations",default_value:false, prompt:"Allowpredictorswithmorethan30categories"})); } varstacked_check=false; if(algorithm=="Regression"){ if(missing=="Multipleimputation") controls.push(form.dropBox({label:"Auxiliaryvariables", types:["Variable:Numeric,Date,Money,Categorical,OrderedCategorical"], name:"formAuxiliaryVariables",required:false,multi:true, prompt:"Additionalvariablestousewhenimputingmissingvalues"})); controls.push(form.comboBox({label:"Correction",alternatives:["None","FalseDiscoveryRate","Bonferroni"],name:"formCorrection", default_value:"None",prompt:"Multiplecomparisonscorrectionappliedwhencomputingp-valuesofpost-hoccomparisons"})); varis_RIA_or_shapley=output=="RelativeImportanceAnalysis"||output=="ShapleyRegression"; varis_Jaccard_or_Correlation=output=="JaccardCoefficient"||output=="Correlation"; if(regressionType=="Linear"&&missing!="Usepartialdata(pairwisecorrelations)"&&missing!="Multipleimputation") controls.push(form.checkBox({label:"Robuststandarderrors",name:"formRobustSE",default_value:false, prompt:"Standarderrorsarerobusttoviolationsofassumptionofconstantvariance"})); if(is_RIA_or_shapley) controls.push(form.checkBox({label:"Absoluteimportancescores",name:"formAbsoluteImportance",default_value:false, prompt:"Showabsoluteinsteadofsignedimportances"})); if(regressionType!="MultinomialLogit"&&(is_RIA_or_shapley||is_Jaccard_or_Correlation||output=="Summary")) controls.push(form.dropBox({label:"Crosstabinteraction",name:"formInteraction",types:["Variable:Numeric,Date,Money,Categorical,OrderedCategorical"], required:false,prompt:"Categoricalvariabletotestforinteractionwithothervariables"})); if(regressionType!=="MultinomialLogit") controls.push(form.numericUpDown({name:"formOutlierProportion",label:"Automatedoutlierremovalpercentage",default_value:0, minimum:0,maximum:49.9,increment:0.1, prompt:"Datapointsremovedandmodelrefittedbasedontheresidualvaluesinthemodelusingthefulldataset"})); stacked_check_box=form.checkBox({label:"Stackdata",name:"formStackedData",default_value:false, prompt:"AllowinputintotheOutcomecontroltobeasinglemultivariableandPredictorstobeasinglegridvariable"}) stacked_check=stacked_check_box.getValue(); controls.push(stacked_check_box); } controls.push(form.numericUpDown({name:"formSeed",label:"Randomseed",default_value:12321,minimum:1,maximum:Number.MAX_SAFE_INTEGER, prompt:"Initializesrandomizationforimputationandcertainalgorithms"})); letallowLargeOutputsCtrl=form.checkBox({label:"Increaseallowedoutputsize", name:"formAllowLargeOutputs",default_value:false, prompt:"Increasethelimitonthemaximumsizeallowedfortheoutputtofixwarningsaboutitbeingtoolarge"}); controls.push(allowLargeOutputsCtrl); if(allowLargeOutputsCtrl.getValue()) controls.push(form.numericUpDown({name:"formMaxOutputSize",label:"Maximumallowedsizeforoutput(MB)",default_value:128,minimum:1,maximum:Number.MAX_SAFE_INTEGER, prompt:"ThemaximumallowedsizeforthereturnedoutputinMB.Verylargeoutputsmayimpactdocumentperformance"})); varoutcome=form.dropBox({label:"Outcome", types:[stacked_check?"VariableSet:BinaryMulti,NominalMulti,OrdinalMulti,NumericMulti":"Variable:Numeric,Date,Money,Categorical,OrderedCategorical"], multi:false, name:"formOutcomeVariable", prompt:"Independenttargetvariabletobepredicted"}); varpredictors=form.dropBox({label:"Predictor(s)", types:[stacked_check?"VariableSet:BinaryGrid,NumericGrid":"Variable:Numeric,Date,Money,Categorical,OrderedCategorical"], name:"formPredictorVariables",multi:stacked_check?false:true, prompt:"Dependentinputvariables"}); controls.unshift(predictors); controls.unshift(outcome); form.setInputControls(controls); if(regressionType==""){ form.setHeading(algorithm); if(form.setObjectInspectorTitle) form.setObjectInspectorTitle(algorithm,algorithm+"outputs"); }else{ form.setHeading(regressionType+""+algorithm); if(form.setObjectInspectorTitle) form.setObjectInspectorTitle(algorithm); } library(flipMultivariates) if(get0("formAllowLargeOutputs",ifnotfound=FALSE)) QAllowLargeResultObject(1e6*get0("formMaxOutputSize")) modelMultivariateStatistics QTechnicalReference>UpdatingandAutomation>AutomationOnlineLibrary ROnlineLibrary UserInterface>CreateRegression UserInterface>Regression Navigationmenu Personaltools Login Namespaces PageDiscussion Variants Views ViewViewsourceHistory More Search Navigation BacktoQhomepageWikihomepageRecentchanges Categories ►InstallingandUpdatingQ►SettingUpData►CreatingAndModifyingTables►Charts►CreatingNewVariables►MultivariateStatistics►UpdatingandAutomation►SharingDataAndResults►Troubleshooting►JavaScript►TestsOfStatisticalSignificance►VideoLibrary Top20CommonProblemsWhenUsingQ ReleaseNotes Tools WhatlinkshereRelatedchangesSpecialpagesPrintableversionPermanentlinkPageinformation