数据分析软件PAST paleontological statistics software package for education and data analysis

PAST-PAlaeontologicalSTatistics

?yvindHammer,D.A.T.Harper,P.D.Ryan

May21,2001

1Introduction

WelcometothePAST!Thisprogramisdesignedasafollow-uptoPALSTAT,anextensivepackagewrittenbyP.D.Ryan,D.A.T.HarperandJ.S.Whalley(Ryanetal.1995).Itincludesanumberoffunctionswhicharecommonlyusedinpalaeon-tologyandpalaeoecology.

Thesedays,anumberoflargeandverygoodstatisticssystemsexist,includingSPSSandExcel.Whyyetanotherstatisticsprogram?

PASTisfree.

PASTistailor-madeforpalaeontology.Thismeansboththatitincludesfunctionswhicharenotfoundinoff-the-shelfprograms(forexamplecladis-tics,ordinationandgeometricalanalysis),andthatitdoesnotincludefunc-tionswhichareoflittleusetopalaeontologistsandthatonlymaketheuserinterfacemoreconfusing.

PASTiseasytouse,andthereforewellsuitedforintroductorycoursesinquantitativepalaeontology.

PASTcomeswithanumberofexampledatasets,casestudiesandexercises,makingitacompleteeducationalpackage.

FurtherexplanationsofmanyofthetechniquesimplementedtogetherwithcasehistoriesarelocatedinHarper(1999).

Ifyouhavequestions,bugreports,suggestionsforimprovementsorothercom-ments,wewouldbehappytohearfromyou.Contactusatohammer@toyen.uio.no.ThePASThomepageis

http://www.toyen.uio.no/~ohammer/past

1

2Installation

ThebasicinstallationofPASTiseasy:Justdownloadthe?le’Past.exe’andputitanywhereonyourharddisk.Double-clickingthe?lewillstarttheprogram.Thedata?lesforthecasestudiescanbedownloadedseparately,ortogetherinthepacked?le’case?les.zip’.This?lemustbeunpackedwithaprogramsuchasWinZip.

Wesuggestyoumakeafoldercalled’past’anywhereonyourharddisk,andputallthe?lesinthisfolder.

Pleasenote:ProblemshavebeenreportedforsomecombinationsofscreenresolutionanddefaultfontsizeinWindows-thelayoutgetsuglyanditmaybenecessaryfortheusertoincreasethesizesofwindowsinordertoseeallthetextandbuttons.Ifthishappens,pleasesetthefontsizeto’Smallfonts’intheScreencontrolpanelinWindows.Weareworkingonsolvingthisproblem.

PASTalsoseemstohaveproblemswithsomeprinters.PostscriptprintersfromHPandTektronixwork?ne.

2

3Enteringandmanipulatingdata

PASThasaspreadsheet-likeuserinterface.Dataareenteredinanarrayofcells,organizedinrows(horizontally)andcolumns(vertically).

Enteringdata

Toinputdatainacell,clickonthecellwiththemouseandtypeinthedata.Thiscanonlybedonewhentheprogramisinthe’Editmode’.Toselecteditmode,ticktheboxabovethearray.Wheneditmodeisoff,thearrayislockedandthedatacannotbechanged.Thecellscanalsobenavigatedusingthearrowkeys.

Anytextcanbeenteredinthecells,butalmostallfunctionswillexpectnum-bers.PleasenotethedecimalpointconventionwhichhasbeenchosenbyWindowsdependinguponyournationality:Acomma(,)oradot(fullstop).’Dotted’datacanbe’commatized’fromtheEditmenu.Absence/presencedataarecodedas0or1,respectively.Anyotherpositivenumberwillbeinterpretedaspresence.

TheconventioninPASTisthatitemsoccupyrows,andvariablescolumns.Threebrachiopodindividualsmightthereforeoccupyrows1,2and3,withtheirlengthsandwidthsincolumnsAandB.Clusteranalysiswillalwaysclusteritems,thatisrows.ForQ-modeanalysisofassociations,samples(sites)shouldthere-forebeenteredinrows,whiletaxa(species)areincolumns.Forswitchingbe-tweenQ-modeandR-mode,rowsandcolumnscaneasilybeinterchangedusingtheTransposeoperation.

Commatize

Convertsallfullstops(’.’)inthedatamatrixtocommas(’,’).Thismaybeneces-saryfortheprogramtoreaddecimalpointscorrectly,dependingonyournational-ity.

Selectingareas

MostoperationsinPASTarecarriedonlyoutontheareaofthearraywhichyouhaveselected(marked).Ifyoutrytorunafunctionwhichexpectsdata,andnoareahasbeenselected,youwillgetanerrormessage.

Arowisselectedbyclickingontherowlabel(leftmostcolumn).

Acolumnisselectedbyclickingonthecolumnlabel(toprow).

Multiplerowsareselectedbyselectingthe?rstrowlabel,thenshift-clicking(clickingwiththeShiftkeydown)ontheadditionalrowlabels.Notethatyoucannot’dragout’multiplerows-thiswillinsteadmovethe?rstrow(seebelow).

3

Multiplecolumnsaresimilarlymarkedbyshift-clickingtheadditionalcol-umnlabels.

Thewholearraycanbeselectedbyclickingtheupperleftcornerofthearray(theemptygreycell)orbychoosing’Selectall’intheEditmenu.

Smallerareaswithinthearraycanbeselectedby’draggingout’thearea,butthisonlyworkswhen’Editmode’isoff.

Renamingrowsandcolumns

WhenPASTstarts,rowsarenumberedfrom1to99andcolumnsarelabelledAtoZ.Foryourownreference,andforproperlabellingofgraphs,youshouldgivetherowsandcolumnsmoredescriptivebutshortnames.Choose’Renamecolumns’or’Renamerows’intheEditmenu.Youmusthaveselectedthewholearray,orasmallerareaasappropriate.

Increasingthesizeofthearray

Bydefault,PASThas99rowsand26columns.Ifyoushouldneedmore,youcanaddrowsorcolumnsbychoosing’Morerows’or’Morecolumns’intheEditmenu.Whenloadinglargedata?les,rowsand/orcolumnsareaddedautomaticallyasneeded.

Movingaroworacolumn

Aroworacolumn(includingitslabel)canbemovedsimplybyclickingonthelabelanddraggingtothenewposition.

Cut,copy,paste

Thecut,copyandpastefunctionsarefoundintheEditmenu.Notethatyoucancut/copydatafromthePASTspreadsheetandpasteintootherprograms,forex-ampleWordandExcel.Likewise,datafromotherprogramscanbepastedintoPAST.

Rememberthatlocalblocksofdata(notallrowsorcolumns)canonlybemarkedwhen’Editmode’isoff.

Allmodulesgivinggraphicoutputhasa’Copygraphic’button.Thiswillplacethegraphicalimageinthepastebufferforpastingintoe.g.WordorCorelDraw.

Remove

Theremovefunction(Editmenu)allowsyoutoremoveselectedrow(s)orcol-umn(s)fromthespreadsheet.Theremovedareaisnotcopiedtothepastebuffer.

4

Grouping(coloring)rows

Selectedrows(datapoints)canbetaggedwithoneofsevenattractivecolorsusingthe’Tagrows’optionintheEditmenu.Eachgroupisalsoassociatedwithasymbol(dot,cross,square,diamond,plus,circle,triangle).Thisisusefulforshowingdifferentgroupsofdataine.g.ternaryandscatterplotsanddendrograms.

Transpose

TheTransposefunction,intheEditmenu,willinterchangerowsandcolumns.ThisisusedforswitchingbetweenRmodeandQmodeinclusteranalysis,principalcomponentsanalysisandseriation.

Loadingandsavingdata

The’Load’functionisfoundintheFilemenu.PASTusesanASCII?leformat,foreasyimportingfromotherprograms(e.g.Word)andeasyeditinginatexteditor.Theformatisasfollows:

.columnlabelcolumnlabelcolumnlabelrowlabeldatadatadatarowlabeldatadatadatarowlabeldatadatadataEmptycells(likethetopleftcell)arecodedwithafullstop(.).Cellsareseparatedbywhitespace,whichmeansthatyoumustneverusespacesinroworcolumnlabels.’OxfordClay’isthusanillegalcolumnlabelwhichwouldconfusetheprogram.

Ifanyrowshavebeenassignedacolorotherthanblack,therowlabelsinthe?lewillstartwithanunderscore,anumberfrom0to6identifyingthecolor(symbol),andanotherunderscore.

The’Insertfrom?le’functionisusefulforconcatenatingdatasets.Theloaded?lewillbeinsertedintoyourexistingspreadsheetattheselectedposition(upperleft).Otherdatasetscanthusbeinsertedbothtotherightofandbelowyourexistingdata.

ReadingandwritingNexus?les

TheNexus?leformatisusedbymanycladisticsprograms.PASTcanreadandwritetheData(charactermatrix)blockoftheNexusformat.Interleaveddataarenotsupported.Also,ifyouhaveperformedaparsimonyanalysisandthe’Parsi-monyanalysis’windowisopen,allshortesttreeswillbewrittentotheNexus?leforfurtherprocessinginotherprograms(e.g.MacCladeorPaup).

5

本文来源：https://www.bwwdw.com/article/wojf.html