> \pJava Excel API v2.6 Ba==h\:#8X@"1Arial1Arial1Arial1Arial + ) , * `JDC,>4$titletitle[alternative]title[translated]author[google(nobreakline)] contributor[author(nobreakline)]"contributor[scopusid(nobreakline)]date[issued]relation[journaltitle]identifier[issn] citationsidentifier[major]subject(nobreakline) publisheridentifier[thesisdegree]contributor[advisor]relation[ispartofseries]relation[index(nobreakline)]typeabstractidentifier[govdoc]identifier[uri]identifier[doi(doi)]identifier[isbn]identifier[ismn]
identifier4lpX XD tǩ\ ) \ 0WComparison of variable selection techniques using tree structured classification method2013Y ĬYtTŐYP YMastert@ǽMaster's Thesis
t Ș <\ | XX0 X XՔ X ) ǔ )@ XǬȘ4t. t )@ X } tt tX䲔 X0 XX ̹D X ̸ pt|ĳ \t XX U1t ɔ t . t| D\ d¸ )@ XǬȘ4 Dt !%t Ř X ɔĳ \ !ĳ| Ĭ` ǌ t . \ ƔȄX4 4lpX X) X\ Ɣ(Projection Pursuit)Ĭ| tǩX X| \ X ɔĳ| ĬX XǬȘ4(Tree)\ д )t.
|8 (X ̸ ǔ @ (X D >0 t Ɣ| X\. t| \T Ȕ \ | > t| tǩX ȹ| XXՔ ƔȄX4@ \ XǬȘ4, d¸| X DP, t |<\ ιt t ǔ |TX @ DPt ଐ \. 10X ̸ X )D P ȩX | DP, X.;CART is the most widely used method for classification. It is easy to understand the structure of classification. However it is very unstable. Random Forest solves this problem using sampling technique. This method presents highly accurate prediction and produces the importance measure of each variable. Projection pursuit classification tree (PPtree) is also one of tree structured classification method and uses projection pursuit method. In each node, PPtree finds a low-dimensional space which shows separated classes. In this situation, we can use projection pursuit coefficients in each node as a measure of importance in variable selection.
In this paper, we compared these tree-structured classification methods in variable selection viewpoints. We also compared these results with the result from the generalized additive model. We apply all these methods to 10 real dataset and compare and analyze the results from each method.~http://dspace.ewha.ac.kr/handle/2015.oak/205364;
http://dcollection.ewha.ac.kr/jsp/common/DcLoOrgPer.jsp?sItemId=000000079413;"@Ce|(!CPrs BKm~ AJlm
,!C
dMbP?_*+%" ,,??U
!
"
#
l>@
Root EntryWorkbookSummaryInformation(DocumentSummaryInformation8