Retail Trade Area Analysis Using the Huff ModelWord文件下载.docx
《Retail Trade Area Analysis Using the Huff ModelWord文件下载.docx》由会员分享,可在线阅读,更多相关《Retail Trade Area Analysis Using the Huff ModelWord文件下载.docx(14页珍藏版)》请在冰豆网上搜索。
![Retail Trade Area Analysis Using the Huff ModelWord文件下载.docx](https://file1.bdocx.com/fileroot1/2023-1/10/836ade0a-a752-49cb-beb8-c6f459e874fd/836ade0a-a752-49cb-beb8-c6f459e874fd1.gif)
TheMicrosoftVirtualEarthPlatformisanintegratedsetofservicesthatcombinesuniquebird’seye,aerial,andsatelliteimagerywithbest-of-breedmapping,locationandsearchfunctionality.Itenablesbusinessestodeliverinnovativesolutionsandbreakthroughcustomerexperiences.Learnmoreat
Introduction
Theconceptofaretailtradeareahasbeenusedbyanalystsandpractitionersinretailsiteevaluationandothermarketstudiesforaverylongtime.Infact,retailtradeareaanalysisandsiteevaluationarecomplementaryprocedures.Retailtradeareaanalysisfocusesonlocatinganddescribingthetargetmarket.Thisknowledgeiscriticalforbothmarketingandmerchandisingpurposes,aswellasforchoosingnewretaillocations.Insiteevaluation,tradeareaanalysisiscombinedwithmanyoperationalrequirementsoftheretailchain(Jones,Simmons1993).
ItismucheasiertoanalyzetradeareasandproducemarketprofilesusingGIS.ThemajorityofGISsoftwareincludesfunctionalityforextractingandaggregatingdataatvariouslevelsofgeography.Asaresult,tradeareaanalysisbecameoneofthemostpopularareasofGISapplicationsinanalyzingbusinessproblems.Themostcommondefinitionofaretailtradeareaisusedforthepurposeofthisarticle.Accordingtothisdefinition,aretailtradeareais“thatarea,typicallyaroundthestore,fromwhichthestorederivesmostofitspatronage”(Lea1998b,p.140).
RetailtradeareaanalysiswasaverypopularthemeduringthetimeBusinessGeographicsmagazinewaspublished(1993-2001).AcoupledozenarticleswerewrittenonthissubjectbyresearchersandGISconsultantsspecializingintheretailsector.Avarietyoftechniquesonhowtodelimitandanalyzetradeareaswerediscussed,alongwiththeiradvantagesanddisadvantages.Thesetechniquesrangefromsimpleones,suchasanapplicationofrings,tomoresophisticated,suchasutilizingprobabilistictradeareasurfaces(Gross1997,Hooper1997,Lea1998a,Peterson1997,Simmons1998).
Alltechniquesrepresenteitherthespatialmonopolyormarketpenetrationapproachestoanalyzingtradeareas(Jones,Simmons1993).Theconcentricringsmethod,drivetime/distancepolygonsorThiessen(Voronoi)polygonsareexamplesofthistypeofapproach.Thesemethodsareeasytoconceptualizeanduse.However,theyassumethatastorehasamonopolyoverthearea–thatallhouseholdsinthetradearearelatetothestoreandnohouseholdsoutsidethetradeareavisitthestore.Oncethetradeareaisdelimitedgeographicallyasaring,Thiessenorothertypeofpolygon,itiseasytoprepareamarketprofilebyextractingandaggregatingdatausingGISsoftware.Althoughthemethodsrepresentingthespatialmonopolyapproacharecommonlyused,theyinvolvealotofsimplificationbecausetheydonotaccountfortheexistenceofcompetingstores.Therefore,theyshouldbeutilizedonlyifnobetteralternativesexist(Lea1998c).
Themarketpenetrationapproachassumesthatthereisaspatialvariationintheproportionofhouseholdsservedbyastoreduetocompetition.ThebestexampleofthistypeofapproachistheHufftradeareamodel.Thetradeareaisconceptualizedasaprobabilitysurface,whichrepresentsthelikelihoodofcustomerpatronage.Thismodelprovidesananswertoabasicquestion:
Whatistheprobabilitythatacustomerwilldecidetoshopataparticularstore,giventhepresenceofcompetingstores?
Thecreationofprobabilitysurfaceisbasedonaspatialinteractionmodelthattakesintoaccountsuchvariablesasdistance,attractivenessandcompetition.Theprobabilitysurfacecanbecontouredtoproduceregionsofpatronageprobability,whichcanthenbefurtherusedasweightsinthepreparationofmarketprofile.
TheintentofthisarticleistoraisetheawarenessoftheHuffmodelwithintheGIScommunity.
IntroductiontotheHuffModel
TheHuffmodelwasintroducedbyDavidHuff
in1963(Huff1963).Itspopularityandlongevitycanbeattributedtoitsconceptualappeal,relativeeaseofuse,andapplicabilitytoawiderangeofproblems,ofwhichpredictingconsumerspatialbehavioristhemostcommonlyknown.Theprobability(Pij)thataconsumerlocatedatiwillchoosetoshopatstorejiscalculatedaccordingtothefollowingformula(Huff2003).
Where:
∙Ajisameasureofattractivenessofstorej,suchassquarefootage
∙Dijisthedistancefromitoj
∙
isanattractivenessparameterestimatedfromempiricalobservations
isthedistancedecayparameterestimatedfromempiricalobservations
∙nisthetotalnumberofstoresincludingstorej.
Thequotientreceivedfromdividing
by
isknownastheperceivedutilityofstorejbyaconsumerlocatedati.The
parameterisanexponenttowhichastore’sattractivenessvalueisraised,andenablestheusertoaccountfornonlinearbehavioroftheattractivenessvariable.The
parametermodelstherateofdecayinthedrawingpowerofthestoreaspotentialcustomersarelocatedfurtherawayfromthestore.Increasingtheexponentwoulddecreasetherelativeinfluenceofastoreonmoredistantcustomers.
Examples
FourexamplesincludedinthisarticleusetheHuffmodelforthefollowingpurposes.
1.Tradeareaanalysisforasinglesiteusingasinglevariableforsiteattractiveness
2.Tradeareaanalysisforasinglesiteusingmultiplevariablesforsiteattractiveness
3.Comparisonofpotentialrevenuefortwosites
4.Modelingamarketscenario–morecomplextradeareaanalysisinvolvingtheuseofcustomerspottingdata,informationonshoppingtripsandmodelcalibration.
Examples1-3aresimpleapplicationsofthemodel.Theydonotinvolveparameterestimationanddonotutilizecustomerspottingdata.TheyarebasedonthedefaultsimplementedinGISsoftwarethatincludestraightlinedistancecalculationandthemosttypicalvaluesforanattractivenessparameter(thevalueof1),andforthedistancedecayparameter(thevalueof2).Evenwiththedefaultvaluesusedforparameters,thismethodissuperiortothemethodsbasedonthespatialmonopolyconcept.
Tradeareaanalysisforasinglesiteusingasinglevariableforsiteattractiveness
Twodatasetswereusedinthisexample:
shoppingcenterswiththeircharacteristics,andsmallcensusunitswithsomerelateddata.Thepurposeofthisexamplewastocreateamarketprofileforasinglemall.TheGrossLeasableArea(GLA)wasusedasanattractivenessvariable.Thepatronageprobabilitysurface(agrid)wascreatedforaselectedmall(Figure1).Apotentialcustomerisassumedtobelocatedateverygridcell.Theprobabilityofacustomerpatronizingaselectedmallispositivelyrelated(directlyproportional)totheattractivenessofthemallandnegativelyrelated(inverselyproportional)tothedistancebetweenthemallandthecustomer,giventhepresenceofallcompetingmalls.
Figure1.CustomerpatronageprobabilitymapforMicmacMall,Halifax-Dartmouth,NovaScotia,usingGrossLeasableArea(GLA)asanattractivenessvariable.(Clickforlargerimage)
Thepatronageprobabilitysurfacewasconvertedtoregionsofprobability.Itispossibletoselectanynumberofregions.Forthisexample,tenregionsofprobabilitywerechosen.TheregionsaredelineatedusingcontoursshownaswhitelinesinFigure1.Thedatawasthenextractedforeachregionfromunderlyingcensuspolygons(Table1).ThenumbersinTable1includeallhouseholdsinthestudyarea.
Table1.Marketprofile(unweighteddata)forregionsofpatronageprobability.(Clickforlargerimage)
Table1wasthensummarizedtoshowthetotalsforeachvariable(Table2).
Table2.Marketprofile(unweighteddata)forregionsofpatronageprobability-summary.(Clickforlargerimage)
Nowthevaluesofprobabilitiescameintoplay.TheywereusedasweightsforscalingdownthenumbersfromTable1,tosimplymakethemmorerealistic.EachofthefirstfourcolumnsinTable1(Population,Dwellings,FamiliesandHouseholds),representingabsolutevalues,wasmultiplied(weighted)bythemidpointvalueofeveryprobabilityregion.Forexample,themidpointvalueforthe0–0.1regionequalsto0.05.Table3presentsmorerealisticmarketareaprofilethatisbasedonweighteddata.ThelasttwocolumnsinTable1werenotweightedbecausetheyrepresentrelativevaluesandthereforearenotincludedinTable3.
Table3.Marketprofile(weighteddata)forregionsofpatronageprobability.(Clickforlargerimage)
Table3wasthensummarizedtoshowthetotalsforeachvariable(Table4).
Table4.Marketprofile(weighteddata)forregionsofpatronageprobability-summary.(Clickforlargerimage)
Thecomparisonoftables2and4allowsforstatingthattheactualnumberofhouseholdspatronizingthismallwillbelessthan10%ofallhouseholdslocatedinthestudyarea.Thisnumberwasdeterminedbycalculatingtheproportionofweightednumberofhouseholds(10,203)inthetotalnumberofhouseholds(110,810).
Tradeareaanalysisforasinglesiteusingmultiplevariablesforsiteattractiveness
ThedifferencebetweenExamples1and2isthatinExample2morethanonevariablewasusedasanattractivenessindex.“Thevalueofthemodeldependsontheabilitytoincorporateanumberofdifferentmeasuresofstoreattractiveness”(Jones,Simmons1993,p.345).Ifmorevariablesareincluded,itiseasiertounderstandavariationinpatronagepatterns.InadditiontotheGLA,inthisexamplethenumberofstoresineachmallandthenumberofparkingspaceswereals