Sie sind auf Seite 1von 2

10/28/2016

RFAQ:HowcanIcalculateMoran'sIinR?

HelptheStatConsultingGroupby

stat

>

>

faq

givingagift

>morans_i.htm

NOTE:Thispagehasbeendelinked.Itisnolongerbeingmaintained,andinformationonthispage
maybeoutofdate.Becausethispagehasbeendelinked,wecannotnotanswerquestions
regardingthispage.
RFAQ:
HowcanIcalculateMoran'sIinR?
Moran'sIisameasureofspatialautocorrelationhowrelatedthevaluesofavariablearebasedonthelocationswherethey
weremeasured.Usingfunctionsintheapelibrary,wecancalculateMoran'sIinR.Todownloadandloadthislibrary,enter
install.packages("ape")andthenlibrary(ape).
Let'slookatanexample.Ourdataset,ozone,containsozonemeasurementsfromthirtytwolocationsintheLosAngeles
areaaggregatedoveronemonth.Thedatasetincludesthestationnumber(Station),thelatitudeandlongitudeofthestation
(LatandLon),andtheaverageofthehighesteighthourdailyaverages(Av8top).Thisdata,andotherspatialdatasets,can
bedownloadedfromtheUniversityofIllinois'SpatialAnalysisLab.Wecanlookatasummaryofourlocationvariablestosee
therangeoflocationsunderconsideration.
ozone<read.table("http://www.ats.ucla.edu/stat/r/faq/ozone.csv",sep=",",header=T)
head(ozone,n=10)
StationAv8topLatLon
1607.22580634.13583117.9236
2695.89919434.17611118.3153
3724.05288533.82361118.1875
4747.18145234.19944118.5347
5756.07661334.06694117.7514
6843.15725833.92917118.2097
7855.20161334.01500118.0597
8874.71774234.06722118.2264
9886.53225834.08333118.1069
10897.54032334.38750118.5347

TocalculateMoran'sI,wewillneedtogenerateamatrixofinversedistanceweights.Inthematrix,entriesforpairsofpoints
thatareclosetogetherarehigherthanforpairsofpointsthatarefarapart.Forsimplicity,wewilltreatthelatitudeand
longitudeasvaluesonaplaneratherthanonasphereourlocationsareclosetogetherandfarfromthepoles.Whenusing
latitudeandlongitudecoordinatesfrommoredistantlocations,it'swisetocalculatedistancesbasedonsphericalcoordinates
(thegeospherepackagecanbeused).
Wecanfirstgenerateadistancematrix,thentakeinverseofthematrixvaluesandreplacethediagonalentrieswithzero:
ozone.dists<as.matrix(dist(cbind(ozone$Lon,ozone$Lat)))
ozone.dists.inv<1/ozone.dists
diag(ozone.dists.inv)<0

ozone.dists.inv[1:5,1:5]
12345
10.0000002.5397952.4461651.6275705.391160
22.5397950.0000002.6670614.5314281.741071
32.4461652.6670610.0000001.9543572.002389
41.6275704.5314281.9543570.0000001.258716
55.3911601.7410712.0023891.2587160.000000

Wehavecreatedamatrixwhereeachoffdiagonalentry[i,j]inthematrixisequalto1/(distancebetweenpointiandpointj).
Notethatthisisjustoneofseveralwaysinwhichwecancalculateaninversedistancematrix.Thisistheformulationused
byStata.InSAS,inversedistancematriceshaveentriesequalto1/(1+distancebetweenpointiandpointj)andthereare
numerousscalingoptionsavailable.
WecannowcalculateMoran'sIusingthecommandMoran.I.

http://www.ats.ucla.edu/stat/r/faq/morans_i.htm

1/2

10/28/2016

RFAQ:HowcanIcalculateMoran'sIinR?

Moran.I(ozone$Av8top,ozone.dists.inv)
$observed
[1]0.2265501
$expected
[1]0.03225806
$sd
[1]0.03431138
$p.value
[1]4.596323e14

Basedontheseresults,wecanrejectthenullhypothesisthatthereiszerospatialautocorrelationpresentinthevariable
Av8topatalpha=.05.
Alternatively,onecoulduseabinarydistancematrix.Thismightbeusedinlookingatnetworkingproblemswheretwopoints
areeitherconnectedornotconnected.Inthisexample,wecanchooseadistancedsuchthatpairsofpointswithdistance
lessthandareconsideredconnectedandpairswithdistancegreaterthandarenotandusethisweightsmatrixtocalculate
Moran'sI.Thecodebelowdoesthisford=.75.
ozone.dists.bin<(ozone.dists>0&ozone.dists<=.75)
Moran.I(ozone$Av8top,ozone.dists.bin)
$observed
[1]0.2045738
$expected
[1]0.03225806
$sd
[1]0.04771132
$p.value
[1]6.910876e07

Inthisexample,weseesimilarresultstothosegeneratedusingtheinverseEuclideandistancematrixandwecanrejectthe
nullhypothesisthatthereiszerospatialautocorrelationpresentinthevariableAv8topatalpha=.05.
Howtocitethispage

Reportanerroronthispageorleaveacomment

Thecontentofthiswebsiteshouldnotbeconstruedasanendorsementofanyparticularwebsite,book,orsoftwareproductbytheUniversityofCalifornia.

IDRE RESEARCH TECHNOLOGY


GROUP

High Performance
Computing
Statistical Computing

GIS and Visualization

ABOUT
2016 UC Regents

CONTACT

NEWS

HighPerformanceComputing

GIS

StatisticalComputing

Hoffman2Cluster

Mapshare

Classes

Hoffman2AccountApplication

Visualization

Conferences

Hoffman2UsageStatistics

3DModeling

ReadingMaterials

UCGridPortal

TechnologySandbox

IDREListserv

UCLAGridPortal

TechSandboxAccess

IDREResources

SharedCluster&Storage

DataCenters

SocialSciencesDataArchive

AboutIDRE

EVENTS

OUR EXPERTS

Terms of Use & Privacy Policy

http://www.ats.ucla.edu/stat/r/faq/morans_i.htm

2/2

Das könnte Ihnen auch gefallen