Please use this identifier to cite or link to this item: http://umt-ir.umt.edu.my:8080/handle/123456789/5652
Full metadata record
DC FieldValueLanguage
dc.contributor.authorMohammad Aizat Basir-
dc.contributor.authorFaudziah Ahmad-
dc.date.accessioned2017-04-11T00:24:04Z-
dc.date.available2017-04-11T00:24:04Z-
dc.date.issued2016-
dc.identifier.citationVol.78(10); 117-124 p.en_US
dc.identifier.issn1279696-
dc.identifier.urihttp://hdl.handle.net/123456789/5652-
dc.description.abstractAttribute selection also known as feature selection is an essential process in data sets that comprise numerous numbers of input attributes. However, finding the optimal combination of algorithms for producing a good set of attributes has remained a challenging task. The aim of this paper is to find a list of an optimal combination search methods and reduction algorithm for attribute selection. The research process involves 2 phases: finding a list of an optimal combination search methods and reduction algorithm. The combination is known as model. Results are in terms of percentage of accuracy and number of selected attributes. Six (6) datasets were used for experiment. The final output is a list of optimal combination search methods and reduction algorithm. The experimental results conducted on public real dataset reveals that the model consistently shows the suitability to perform good classification task on the selected dataset. Significant improvement in accuracy and optimal number of attribute selection is achieved with a list of combination algorithms used.en_US
dc.language.isoenen_US
dc.publisherJurnal Teknologien_US
dc.subjectPemilihan atributen_US
dc.subjectalgoritma penguranganen_US
dc.subjectkaedah carianen_US
dc.subjectpengelasanen_US
dc.subjectAttribute selectionen_US
dc.subjectreduction algorithmen_US
dc.subjectsearch methodsen_US
dc.subjectclassificationen_US
dc.titleAttribute Selection Model For Optimal Local Search And Global Searchen_US
dc.typeArticleen_US
Appears in Collections:Journal Articles



Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.