下载
加入VIP
  • 专属下载特权
  • 现金文档折扣购买
  • VIP免费专区
  • 千万文档免费下载

上传资料

关闭

关闭

关闭

封号提示

内容

首页 GWR4_Manual

GWR4_Manual.pdf

GWR4_Manual

damofeixuec
2013-01-04 0人阅读 举报 0 0 暂无简介

简介:本文档为《GWR4_Manualpdf》,可适用于人文社科领域

GWRUserManualGWRWindowsApplicationforGeographicallyWeightedRegressionModellingTomokiNakayaUpdatedMayJuneGWRDevelopmentTeamTomokiNakaya(DepartmentofGeography,RitsumeikanUniversity)MartinCharlton,PaulLewis(NCG,NationalUniversityofIreland)StewartFotheringham(SchoolofGeographyGeosciences,UniversityofStAndrews)ChrisBrunsdon(DepartmentofGeography,UniversityofLiverpool)ContentsIntroductionWhatisGWRNewfeaturescomparedtoGWRxNotesforuseofGWRInstallationUninstallationStartingtheprogram,Exitingtheprogram,andTabdesignFivestepsinGWRcalibrationStep:TheDataTabDatapreparationOperationsinthedatatabpageStep:TheModelTabBasicoperations:usinganexampleofGaussianGWRSemiparametricGWRExtensionsofGWR:GWGLMGeographicallyweightedPoissonregression(GWPR)Geographicallyweightedlogisticregression(GWLR)ModellingOptionsStandardisationGeographicalvariabilitytestLtoGGtoLvariableselectionStep:TheKernelTabPossiblefixedandadaptivekernelfunctionsforgeographicalweightingBandwidthselectionroutinesSelectioncriteriaStep:TheOutputTabSessionControlFileCommonoutputfilesThe“Predictionatnonregressionpoints”optionStep:TheExecuteTabTheExecutebuttonFieldsinalistwiseoutputHandlingasessionandbatchmode(optional)ExampleoutputReferencesIntroductionWhatisGWRGWRisanewreleaseofaMicrosoftWindowsbasedapplicationsoftwareforcalibratinggeographicallyweightedregression(GWR)models,whichcanbeusedtoexploregeographicallyvaryingrelationshipsbetweendependentresponsevariablesandindependentexplanatoryvariablesAGWRmodelcanbeconsideredatypeofregressionmodelwithgeographicallyvaryingparametersAconventionalGWRisdescribedbytheequationkiikiikixvuy,),(,whereiy,ikx,,andiare,respectively,dependentvariable,kthindependentvariable,andtheGaussianerroratlocationi(iivu,)isthexycoordinateoftheithlocationandcoefficients),(iikvuarevaryingconditionalsonthelocationSuchmodellingislikelytoattainhigherperformancethantraditionalregressionmodels,andreadingthecoefficientscanleadtoanewinterpretationofthephenomenaunderstudyAnimportantextensionofGWRisitssemiparametricformationbymixinggloballyfixedandgeographicallyvaryingcoefficientskilillikiikizxvuy,,),(whereilz,isthelthindependentvariablewithafixedcoefficientlSuchamodelmayreducethemodelcomplexitiesandenhanceitspredictableperformanceUsingtheframeworkofgeographicallyweightedgeneralisedlinearmodelling(GWGLM),logisticandPoissonregressionmodelswithgeographicallyvaryingcoefficientsarealsopopularforbinaryorcountdatamodellingGWRenablesthefittingofsuchGWRandGWGLMmodelswiththeirsemiparametricformations,associatedstatisticaltests,andmodelselectionsbyuserdefineddataandmodelsettingsLocaltermsGlobaltermsMainfeatures()SemiparametricGWRAsnotedabove,amostremarkablefeatureofthisreleaseisthefunctiontofitsemiparametricGWRmodels,whichallowyoutomixgloballyfixedtermsandlocallyvaryingtermsofexplanatoryvariablessimultaneouslyThefunctioncanbeappliedtopopulartypesofgeneralizedlinearmodellingincludingGaussian,Poisson,andlogisticregressionsUsingthesemiparametricmodellingscheme,anewstatisticaltestofgeographicalvariabilityongeographicallyvaryingcoefficientsisenabledItisalsopossibletousevariableselectionroutinesbywhichvariablesareautomaticallyselectedaseitherfixedorvaryingtermsbyrecursivemodelcomparisons()InterfaceAtabbedinterfacehasbeenintroducedtoenablemodellingsessionstointuitivelyproceedinastepbystepmannerDatasetsandgeographicallylistwiseresultscanbeviewedinseparatespreadsheetlikewindowsSeveralpopularfiletypescanbeusedasinputdatafiles(space,comma,tabseparatedtext,anddbaseIVformats)Inaddition,ArealkeyfieldcanbeintegratedintotheoutputofGWRmodelling,enablingyoutojoinyouroutputCSVfiletoaGISattributetableviathekeyfieldformappingtheresultinaGISenvironmentGWRcanbealsousedbyabatchmodewithouttheWindowsinterface()RequirementsGWRrunsonWindowsVistaorWindowsenvironmentswiththeNETFrameworkThemaximumsizeofdataisdependentonyourlocalmachineenvironmentGWRdynamicallyallocatesmemoryforlargematrices(nbyn,wherenisthenumberofregressionpoints)evenforconventionalGWRmodelsThususingaPChavingrelativelylargememorysize(equaltoorlargerthanGB)forrunningGWRisrecommendedIfthePChasmulticoreprocessors,GWRautomaticallyusesmultithreadingroutinestospeedupthecomputationNotesforuseofGWR(i)GWRiscopyrightedbytheGWRdevelopmentteam(ii)GWRcanbefreelydistributedandusedforacademicandnonprofitpurposesThedevelopersofGWRarenotresponsibleforanydifficultiesthatusersofthesoftwaremayencounter(iii)WhenanyresultsusingGWRarepublished,theauthor(s)shouldclearlystatethatGWRwasusedInstallationUninstallation<HowtoinstallGWR>DownloadtheGWRinstaller,GWRsetupmsi,andthendoubleclicktheicon<NETFramework>GWRworksonlyinMSWindowsenvironmentsthathavetheNETFrameworkinstalledIfyourPCdoesnothaveitinstalled,amessagewillpopupsuggestingthatyoudownloadtheNetFrameworkClientProfilefromaMicrosoftwebsiteAftersettinguptheNETFramework,youcanthenretryinstallationofGWR<Whentheinstallerstarts…>FollowtheinstructionstoselecttheGWRinstallationfolderandusersIftheinstallationissuccessful,ashortcuttotheprogramwillappearonyourdesktopandintheGWRprogramgroupYoumayaccesstheprogrambyclickingthisshortcut<Touninstall>TouninstallGWRfromyourlocalenvironment,doubleclickthesetupfile,GWRsetupmsi,againandselect“delete”Alternatively,youcanusethe“UninstallPrograms”utilityintheControlPanelgroupinVistaandWindowsenvironmentsStartingtheprogram,Exitingtheprogram,andTabdesign<Startingtheprogram>Tostarttheprogram,doubleclicktheGWRshortcuticononthedesktop,orselectitfromtheGWRprogramgroup<Tabdesign>EnsurethattherearefivetabslabelledSteptoStepClickatablabeltomovetothecorrespondingtabpageThefirsttabpagewhentheprogramstartsis“Step:Data>”Figure:GWRstartupscreen<Exitingtheprogram>Toexittheprogram,select“Quit(Q)”ontheFilemenu(alternatively,youcanpresstheAltand“F”keyssimultaneouslyandthenpressthe“Q”key),oryoucanclicktheclosebuttoninthetoprightcornerofthewindowTablabelsFivestepsinGWRcalibrationWeregardasessionastheoverallprocessbywhichsettingsareusedtocalibrateaGWRmodelTobuildandproceedthroughyourownsession,youcangenerallyfollowthefollowingfivesteps(Figure)EachstepisseparatedintoatabbedpageofthesoftwareAsdescribedbefore,tomovetoadifferentpage,clickthecorrespondingtablabelintheupperpartofthewindowFigure:StepsinaGWRmodellingsessionwithrespecttothefivetabsStep:Execute>ExecutethesessiontocomparenecessarycalculationsandreadresultsStep:Output>SpecifyfilenamesforthefilesstoringthemodellingresultsStep:Kernel>ChooseageographickerneltypeanditsbandwidthsizeAutomatedoptimisationofbandwidthsizeisalsoavailableStep:Model>SpecifyoneregressiontypeandthevariablesettingsneededforGWRmodellingStep:Data>StartyoursessionbygivingitatitlethenopenyourdatafileStep:TheDataTabDatapreparation<WhatfieldsdoIhavetoprepareinmydataset>TocalibrateaGWRmodel,youmustprepareatabulardatasetthatcontainsfieldsofdependentandindependentvariables,andxycoordinatesEveryvariableshouldconsistofnumericvalues,exceptforArealkey,asanidentifierofobservationArealkeyistreatedasastringfieldinGWR<Coordinates>Bothprojectedandlatitudelongitude(latlon)decimaldegreescoordinatescanbeusedasxycoordinatesinGWRHowever,projectedcoordinatesaresuperiortodecimaldegreesintermsofcomputingtime<Possibledataformats>GWRcanopendatafilesintextformatdelimitedbyspace,comma,ortabaswellasfilesindbaseIVformat(*dbf)ThemostpopulartypeisCSV(atextformatthatusesthecommadelimiter)<dbaseIVfiles>SincetheshapefilespatialdataformatacommonGISfileformatfromESRIIncusesdbaseIVasanattributedatabaselinkingwithgeographicobjects,thefunctiontoreaddbasefileprovidesaneasywaytousesuchaGISdatafileinGWRHowever,thereisonecaveatforthisfunction:InthecaseofdbaseIV,thelengthofthefilenamemustnotexceedcharacters,duetotherestrictionondatabaseconnectionusedinthissoftwareSomeexamplesofreadableandunreadablefilenamesareshownbelow:Readablefilenames:tokyomdbf,tokyodatdbfUnreadablefilenames:dublindatadbf,irelandmortalitydbf<Fieldnames>Inthecaseoftextformats,thefirstline(row)ofthedatafilemustbethelistoffieldnames(AdbaseIVfileautomaticallyhasthefieldnamelists)<Samplesoftextbaseddatafiles>TwosamplesofdatafilesusingspaceandcommadelimitersareshownbelowTable:SampleoftextdatafileforGWR:Tokyomortalitydata(Spacedelimitedtextfile)Note:onlythefirsttenrecordsareshownTable:SampleoftextdatafileforGWR:Georgiadata(Commadelimitedtextfile:CSV)Note:onlythefirsttwentyrecordsareshown<Missingvalues>GWRdoesnothavefunctionsforhandlingmissingcodesRecordshavingblankitemsinthedependentvariablefieldareskippedforthemodelfittingInthecasesthatthereareblankitemsforindependentvariables,themodelfittingfailsAvoidincludingmissingvaluesforyourdatasetbyexcludingsuchdataorestimatedmissingvaluesbeforeGWRmodellingOperationsinthedatatabpageFigure:Samplescreenshotfor“Step:Data”Thefollowingistherecommendedproceduretofollowonthe“Data”tabpage:()Createasessiontitlethatreflectsyourmodellingintentionanddataset()Inputthefilenameofyourdatasetinthistextboxbyclickingtherightmost“Browse”buttontoopenafiledialogbox()Selecttheappropriateformatordelimiterthatseparateseachvalueinarowofatextdatafile()Clickthe“Open”buttonBoth“Numberoffields”and“Numberofareas”willautomaticallyappear(Note:“Numberofareas”referstothenumberofdatarecordsobservationsinthedatafile)()Youcanviewyourdatasetinaspreadsheetlikewindowbyclickingthe“Viewdata”buttonThenumberofrows(lines)inthegriddedviewcanbecontrolledusingtheoptionstotheleftofthebutton(defaultisfirstlines)()()()()()Figure:DataviewwindowStep:TheModelTabIfthedatafileyouspecifiedonthe“Data”tabpageissuccessfullyopened,fieldnameswillappearinthe“Variable(Field)list”listboxinthemiddleofthe“Model”tabpageIfthereisnofieldnameorthelistingsinthelistboxareinsufficient,returntothe“Data”tabpage,thenpressthe“Open”buttonafterselectingaproperdatafileandformatdelimiteroptionFigure:ScreenshotofthemodeltabpagejustafteropeningadatafileIfyouclickthefieldtitle,therowswillbesortedbyfieldvalueBasicoperations:usinganexampleofGaussianGWR<GaussianGWRmodel>Itisessentialtoensurethatappropriateoptionsareselectedforyourmodelstructureandvariablesinthis“Model”tabpageAconventional(Gaussian)GWRmodelisdescribedaskiikiikixvuy,),(,whereiy,ikx,,andiare,respectively,dependentvariable,kthindependentvariable,andtheGaussianerroratthelocationi(iivu,)isthexycoordinateoftheithlocationandcoefficients),(iikvuarevaryingconditionalonthelocationUsually,thefirstvariableisconstantbysetting,ix,afterwhich),(iivubecomesageographicallyvarying“intercept”term<FittingaGaussianGWRmodel>TofitaGaussianGWRmodel,atleastthreekindsofoperationsshouldbedone:()<variablesettings>:Byusingthe“<”andthe“>”buttons,selectandmovetheappropriatefieldofdependentvariabletothe“dependentvariable”textboxInasimilarfashion,movethefieldsofindependentvariablesto“Local”fromthevariable(field)listTheitem“Intercept”isincludedinthe“Local”listboxforgeographicallyvaryingtermsasadefaultconditionBymovingthe“intercept”listitemtothevariable(field)list,theinterceptcanbefixedaszerointhemodel(nointerceptmodel)However,inmostcases,itisrecommendedthat“intercept”notberemovedfromyourmodel()<coordinatespecification>:Similarly,selectandmoveappropriatefieldsfromthevariable(fields)listto“xcoordinatevariable”and“ycoordinatevariable”thenselecttheappropriatecoordinateoption(either“projected”or“spherical”)“Projected”istypicallyusedforcoordinatesprojectedontoanorthogonaltwodimensionalspace,suchasUTMcoordinates“Spherical”shouldbeselectedifxycoordinatesarestoredinaLatLonformat(decimaldegrees)Thesphericaldistanceismoreaccurateifthegeographicextentofthestudyareaiswide(forexample,continentalscale)However,sphericaldistancecalculationiscomputationallymoreexpensivethanEuclidiandistancecalculationappliedto“Projected”coordinatesFormostcityandregionalscaleapplications,projectedcoordinatesaresuitableintermsofthebalancebetweenaccuracyandcomputationalload()<modeltype>Selectthe“Gaussian”optioninthemodeltypepanel<ExampleofGaussianGWR>ThefollowingscreenshotshowsanexampleofaGaussianGWRmodelusingtheGeorgiasampledatawiththefollowingspecifications:PctBatch(,)(,)PctRural(,)PctPov(,)PctBlackiiiiiiiiiiiiiXYXYXYXYwhereXiandYiareprojectedxycoordinatesinthisexampleFigure:ScreenshotofthemodeltabpageforasimpleGaussianGWR()()()()()<IDKeyoption>IfyouhavealocationalIDfield(suchasplacenamesorregionalcodes),youcanincludeitforlateruseinyourresultsUsingthisIDkey,youcanjoinyourresultingoutputfilescontaininglocalestimatestoothertables(suchasGISattributetables)Figure:IDkeyoptionSemiparametricGWR<SemiparametricGWR>AsemiparametricGaussianGWRmodelisdescribedaskilillikiikizxvuy,,),(whereilz,isthelthindependentvariablewithafixedcoefficientlThus,themodelmixesgeographicallylocalandglobaltermsThiskindofmodelhasseveralaliases,suchas“mixedmodel”and“partiallinearmodel”TheintercepttermisusuallyspecifiedasavaryingtermsinceothervaryingcoefficientsoftencauseavariationintheinterceptHowever,itisalsopossibletoassigntheinterceptasafixedtermintheinterface<Variablesettingsforsemiparametricmodels>Tofitasemiparametricmodel,mostoftheoperationsneededarethesameasthoseusedinthecaseoftraditionalGWRmodelsOneadditionalthingistospecifytheglobaltermbymovingindependentvariablesforglobaltermstothe“Global”boxLocaltermsGlobalterms<ExampleofsemiparametricGaussianGWR>ThefollowingequationisanexampleofasemiparametricGaussianGWRmodelusingtheGeorgiasampledatawiththefollowingspecifications:PctBatch(,)(,)PctRural(,)PctPov(,)PctBlackPctEldPctFBiiiiiiiiiiiiiiiXYXYXYXY

用户评价(0)

关闭

新课改视野下建构高中语文教学实验成果报告(32KB)

抱歉,积分不足下载失败,请稍后再试!

提示

试读已结束,如需要继续阅读或者下载,敬请购买!

文档小程序码

使用微信“扫一扫”扫码寻找文档

1

打开微信

2

扫描小程序码

3

发布寻找信息

4

等待寻找结果

我知道了
评分:

/40

GWR4_Manual

VIP

在线
客服

免费
邮箱

爱问共享资料服务号

扫描关注领取更多福利