\contentslabel
[\thecontentslabel]1.25cm \titlerule*[.5pc]. \thecontentspage\contentslabel[\thecontentslabel]1.25cm \thecontentspage[]\contentslabel[\thecontentslabel]1.25cm \titlerule*[.5pc]. \thecontentspage[]Thenextstepwastobuildthedatacube,atthispointyoucandecideifyouwanttoprocessyourimagesindendentlyorallofthem.Theidealhereistoinputalloftheminadatacube,sotheoutputclutersrelateinformationfromallthewavelenghtsandtheregionscoveredbythemcanbeinterpretedmoreeasily.Nowifyouchoosetocreateanimagecube(justappendtheimagearraysinoneFITSfile)itisposiblethatyouyimageshaveadifferentconversionbetweentheirworldcoordinatesystemtopixel,sohavetomakesureallofyourimagesareprojectedwithonlyoneconversion,thismeanthatyouhavetoreprojectthemtoacommonWCS.Well,whatIwrotebeforeitisabriefsummaryofwhatIdid,butI’msurethatyoucanfindabetterwaytodoyourowndatapre-processingbutherearesomethingsthatyoushouldconsider:
-
Createamethondasgeneralaspossible,withinputparameterthatcanbeadaptedtoanykindofdata,thiswillsaveyoualotofworkinthefuture
-
Understandfirstyouralgorithm,howthedataisgoingtobeprocessedanddesignthebestwaytoinputyourdata
-
Accomodateyourdataaccordingtothetypeofattributesthatthealgorithmcanhandle
-
Considerthesizeofyourdataset,ifit’shugeyourprogrammayneverend
-
Findoutofyouralgorithmcanworkwithhighdimensionaldata(multi-wavelenght),becauseifnot,youwon’tbeabletoinputdatacubes
-
Findoutifyourselectedclusteringalgorithmsisabletofindclustersofirregularshapes,thiswillhelpyoutodevicethebestwaytoaccomodateyourpatterns
-
Handleoutliers,ifyouidentifythem,knowwheretheyare,trytoeliminatethemasmuchaspossible,wedon’twantthemmessingwithourclusters
-
IncasethatyoucomeupwithanartfulmathematicalmethodlikePCAtoreducedimensionality,makesurethatwhatyouinputcanlatermakesensewhenisclustered,becuaseyouwillbeworkinginanotherspace
-
Remeberthatthemostimportantgoalistofindhiddenknowledgetherefore,youmustknowyoutovisualizeandinterpretyourresults
-
Forthelet’scallitastronomyimageprocessing,makesurethatyourdataisscientificallyaprovedaskpeoplearoundyou.
ThissectionisexplainedatlenghtintheGitHubpage,thereyouwillfindmycodesandsomehelpfullinks,https://github.com/LaurethTeX/Clustering/blob/master/Preprocessing.md