a b s t r a c tA recentseriesofpape

a b s t r a c t
A recentseriesofpapersbyCharlesT.Perrettiandcollaboratorshaveshownthatnonparametric
forecasting methodscanoutperformparametricmethodsinnoisynonlinearsystems.Suchasituation
can arisebecauseoftwomainreasons:theinstabilityofparametricinferenceproceduresinchaotic
systemswhichcanleadtobiasedparameterestimates,andthediscrepancybetweentherealsystem
dynamics andthemodeledone,aproblemthatPerrettiandcollaboratorscall “the truemodelmyth”.
Should ecologistsgoonusingthedemandingparametricmachinerywhentryingtoforecastthe
dynamics ofcomplexecosystems?Orshouldtheyrelyontheelegantnonparametricapproachthat
appears sopromising?Itwillbeherearguedthatecologicalforecastingbasedonparametricmodels
presents twokeycomparativeadvantagesovernonparametricapproaches.First,thelikelihoodof
parametric forecastingfailurecanbediagnosedthankstosimpleBayesianmodelcheckingprocedures.
Second, whenparametricforecastingisdiagnosedtobereliable,forecastinguncertaintycanbe
estimated onvirtualdatageneratedwiththe fitted todataparametricmodel.Incontrast,nonparametric
techniquesprovideforecastswithunknownreliability.Thisargumentationisillustratedwiththesimple
theta-logistic modelthatwaspreviouslyusedbyPerrettiandcollaboratorstomaketheirpoint.Itshould
convince ecologiststosticktostandardparametricapproaches,untilmethodshavebeendevelopedto
assess thereliabilityofnonparametricforecasting.

ผลลัพธ์ (ไทย) 2:

นามธรรม
recentseriesofpapersbyCharlesT.Perrettiandcollaboratorshaveshownthatnonparametric
methodscanoutperformparametricmethodsinnoisynonlinearsystems.Suchasituation
การพยากรณ์สามารถ andthemodeledone, aproblemthatPerrettiandcollaboratorscall ว่า "truemodelmyth". ควร ecologistsgoonusingthedemandingparametricmachinerywhentryingtoforecastthe การเปลี่ยนแปลง ofcomplexecosystems? Orshouldtheyrelyontheelegantnonparametricapproachthat ปรากฏ sopromising? Itwillbeherearguedthatecologicalforecastingbasedonparametricmodels นำเสนอ twokeycomparativeadvantagesovernonparametricapproaches.First, thelikelihoodof forecastingfailurecanbediagnosedthankstosimpleBayesianmodelcheckingprocedures พารา. ประการที่สอง whenparametricforecastingisdiagnosedtobereliable, forecastinguncertaintycanbe ประมาณ onvirtualdatageneratedwiththe ติดตั้ง modelthatwaspreviouslyusedbyPerrettiandcollaboratorstomaketheirpoint.Itshould โน้มน้าว ecologiststosticktostandardparametricapproaches, untilmethodshavebeendevelopedto ประเมิน thereliabilityofnonparametricforecasting

ผลลัพธ์ (ไทย) 3:

B S T R A C T
เป็น recentseriesofpapersbycharlest . perrettiandcollaboratorshaveshownthatnonparametric

สามารถพยากรณ์ methodscanoutperformparametricmethodsinnoisynonlinearsystems . suchasituation arisebecauseoftwomainreasons : theinstabilityofparametricinferenceproceduresinchaotic systemswhichcanleadtobiasedparameterestimates andthemodeledone andthediscrepancybetweentherealsystem

, พลศาสตร์aproblemthatperrettiandcollaboratorscall " truemodelmyth "

ควร ecologistsgoonusingthedemandingparametricmachinerywhentryingtoforecastthe พลวัต ofcomplexecosystems ? orshouldtheyrelyontheelegantnonparametricapproachthat
ปรากฏ sopromising ? itwillbeherearguedthatecologicalforecastingbasedonparametricmodels
ของขวัญ twokeycomparativeadvantagesovernonparametricapproaches ครั้งแรกthelikelihoodof
พารา forecastingfailurecanbediagnosedthankstosimplebayesianmodelcheckingprocedures .
2 whenparametricforecastingisdiagnosedtobereliable forecastinguncertaintycanbe
, onvirtualdatageneratedwiththe ประมาณพอดี todataparametricmodel . และ thisargumentationisillustratedwiththesimple
3
techniquesprovideforecastswithunknownreliability .ที โลจิสติก modelthatwaspreviouslyusedbyperrettiandcollaboratorstomaketheirpoint . ทั้งนี้ให้ ecologiststosticktostandardparametricapproaches
,
untilmethodshavebeendevelopedto ประเมิน thereliabilityofnonparametricforecasting .