Treść książki
Przejdź do opcji czytnikaPrzejdź do nawigacjiPrzejdź do informacjiPrzejdź do stopki
JerzyTchórzewski,TomaszKania
pressedasthedistancebetweenthemostdistantneighbors,i.e.posi-
tionedelementsofsets,andthismethodworkswellwhentheobjects
formseparateclusters,whileinthecaseofclustersarrangedinanelon-
gatedformwithchaincharacteristics,thismethoddoesnotbringfavor-
ableresults;
−themethodofaverageconnections(UnweightedPair-GroupMeth-
odsUsingArithmeticAverages),whichwasdevelopedinordertolimit
theinfluenceofstronglydiverging,i.e.extremevalues,onthecriterion
ofgroupconnection.Inthemeanlinkagemethod,thedistanceiscal-
culatedbasedonthearithmeticmeanofthedistanceofallmembers
belongingtodifferentclusters.Theresultingclustergroupstendtohave
verylowvariabilitywithinonegroup.Themethodisapplicableboth
whenclusterstaketheformofelongatedclustersandinthecaseof
objectsclusteredinseparatedgroups;
−theweightedpalr-groupmethodusingarithmeticaverages,whichis
similartothestandardmethodofaverageconnections,whichwasde-
velopedbytakingintoaccountthesizeoftheclusters,i.e.thenumber
ofobjectscontainedinthem.Thismethodisusedwhenthereisapo-
tentiallysignificantdiscrepancybetweenthesizeoftheanalyzedsets;
−thecenterofgravitymethod,alsoknownasthecentroidmethodor
unweightedpair-groupofusingthecentroidaveragemethods,inwhich
thedistancebetweenapairofclustersisrepresentedasthevalueof
thesegmentjoiningtheircentersofgravitycalculatedonthebasisof
theaveragedvaluesofthefeaturesoftheelementswithinthecluster.
Theaveragevaluesarecalledcentroid.Thistypeofmethodisvery
similartothemethodofdeterminingthecenterofgravityonamapus-
inggeographiccoordinates;
−themedianclusteringmethod,alsoknownastheweightedgroup
centroidmethod,issimilartothemethodofmeanconnectionswiththe
difference,however,thatpossibledifferencesbetweenthesizeofthe
clustersaretakenintoaccount,thankstowhich,inthecaseofclusters
withhighlydifferentiatedcardinality,thedistancebetweentheseclus-
terswillbebetterreflected;
−theward'serrorsumofsquaresmethod,alsoknownastheincrease
insumofsquaresmethod,usesanalysisofvariancetoestimatethe
distancebetweenclusters.Thismethodisaimedatminimizingthesum
ofsquaresofdeviationsinclustersandusestheErrorSumofSquares
(ESS)asameasureofdifferentiationwiththesearchfortheminimum
16
