资源描述:
《Classification techniques for time series and functional data》由会员上传分享,免费在线阅读,更多相关内容在学术论文-天天文库。
1、ClassificationTechniquesforTimeSeriesandFunctionalDataInternetVersionDavidCasadodeLucasDoctoralThesisAdvisors:AndresM.AlonsoandJuanJ.Romo12thJuly2010Somecalculus-tricksarequiteeasy.Someareenormouslydicult.Thefoolswhowritethetextbooksofadvancedmathematics
2、andtheyaremostl
3、ycleverfools
4、seldomtakethetroubletoshowyouhoweasytheeasycalculationsare.Onthecontrary,theyseemtodesiretoimpressyouwiththeirtremendousclevernessbygoingaboutitinthemostdicultway.Beingmyselfaremarkablystupidfellow,Ihavehadtounteachmyselfthedicul-ties,andnowbegtopresenttomy
5、fellowfoolsthepartsthatarenothard.Masterthesethoroughly,andtherestwillfollow.Whatonefoolcando,anothercan.SilvanusP.Thompson(fromtheprologueofCalculusMadeEasy)Tomyfamily,fortheabsolutecondencetheyhavealwaysplacedinme(evenwhentheydidnotunderstandwhymy
6、our
7、salaryandholidays
8、werenotlikethoseteachersusuallyhave").iPrefaceThefollowingbodyofwork1ispartoftheresearchthatIhavedone,incollaborationwithmyadvisors,overseveralacademicyearsattheUniversidadCarlosIIIdeMadrid.IamverypleasedtohavehadtheopportunitytohelpwiththeteachingtasksattheDepartamentod
9、eEstadstica.Thisdocumentisadigitalversion
10、bycommonconsentofmyadvisors
11、ofmydoctoralthesis,defendedon12thJuly2010.Somemisprintshavebeenremovedandfewexplanations,remarks,equationsandpictureshavebeenincluded.Themarginshavebeenreducedandtheinterlinespacinghasbeensettooneanda
12、half,insteadofdouble.Asaresult,alotlesspageswereneeded.Severalappendiceswithauxiliartheoryhavebeenincluded(Iamsolelyresponsibleforthesesections).Theycontainwell-knowntheoryrelatedtothemathematicsused
13、inamoreorlessdirectway
14、inthepreviouschapters.Manystatisticiansarenotmath
15、ematicians;forthisreason,theindexhasbeenexpandedtomaketheseappendicesmoreaccessible.Readerswithenoughbackgroundknowledgewillnotneedtoconsultthem.Thistextisorganizedasfollows:therstchapterbeginswiththebasicmathematicalnota-tion,followedbythemaindenitionsandclassicalresul
16、ts,andconcludeswithgeneralideasonthestatisticalproblemofclassicationareprovided,justbeforetheex