Please use this identifier to cite or link to this item: http://theses.ncl.ac.uk/jspui/handle/10443/5945
Full metadata record
DC FieldValueLanguage
dc.contributor.authorKoc, Kenan-
dc.date.accessioned2023-11-27T10:09:53Z-
dc.date.available2023-11-27T10:09:53Z-
dc.date.issued2022-
dc.identifier.urihttp://hdl.handle.net/10443/5945-
dc.descriptionPhD Thesisen_US
dc.description.abstractAbilities such as visualization and interaction play essential roles in data mining since they can help people grasp and explore information more easily through their ability to bring out complex and multivariate patterns in data. The research presented in this thesis exploits and demonstrates the powerful combination of visual representation, domain knowledge and machine learning techniques to support challenges related to forming balanced groups. Teamwork is of substantial interest in academia and industry since interpersonal skills count in modern society. A team can, for example, be defined as a group gathered around a common project. Education is one of the domains in which group studies are important, and research studies are done to increase the effectiveness of group studies. In this case, forming appropriate groups for tasks at hand may be overwhelming for educators, as several factors may affect the quality of the group output. The current research supports educators in the group formation process and explores how to form groups systematically with less bias. In this thesis, a holistic framework called GroupVis, is presented in which exploration, clustering and grouping are considered user workflow aspects of group formation. In the GroupVis, each of these aspects is designed as a module, and each module contains visualization and computational methods within itself. The framework is designed with a top-down flow, where the result of the higher modules acts as input to the lower modules. As part of the GroupVis research a novel glyph was designed and evaluated, as a method that supports the comparison of balanced patterns in multivariate data. The three main modules of the GroupVis support group formation with functions such as analysing the data attribute field, exploring the cluster field visually with different settings, creating desired type groups, and interactively examining and modifying the resulting groups. The effectiveness of the framework and its modules is evaluated through semi-structured interviews with target users as well as through a heuristic survey with domain experts in education. The approaches presented here were designed and developed to be practical and applicable for the formation of groups in a wide range of domains, with the educational setting being one of these domains where we recognized their usefulness.en_US
dc.language.isoenen_US
dc.publisherNewcastle Universityen_US
dc.titleInformation visualization approach to form balanced groupsen_US
dc.typeThesisen_US
Appears in Collections:School of Computing

Files in This Item:
File Description SizeFormat 
KocK2022.pdfThesis5.24 MBAdobe PDFView/Open
dspacelicence.pdfLicence43.82 kBAdobe PDFView/Open


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