Please use this identifier to cite or link to this item:
Title: An Efficient Visual Approach For Automatic Clustering And Validation
Researcher: Puniethaa Prabhu
Guide(s): Duraiswamy K
Keywords: algorithms
data mining
Upload Date: 29-Sep-2014
University: Anna University
Completed Date: n.d.
Abstract: Clustering or exploratory data analysis is a widely applied newlineunsupervised technique in the data mining domain The major concern of the newlinedomain is how the observed data can be categorized into meaningful newlinestructures However most of the existing clustering algorithms are not newlineadequate in dealing with arbitrarily shaped distribution of data such as data newlinesets of extremely large volume data visualization and data sets of highdimensional newlinefeatures The key limitations of Indexbased and Statisticalbased newlinecluster validation methods are that making unrealistic distributional newlineassumptions of the data and incurring high computational cost in cluster newlineanalysis which prevents the clustering algorithms from being efficiently used newlinein practiceThe count of clusters is considered as a key factor for clustering newlineoperations in most of clustering algorithms Therefore the quality of the newlineresultant clusters mainly depends on the assessment of cluster number The newlineClustering of unlabeled data set faces certain critical issues such as assessing newlinecluster tendency ie determining the number of clusters prior to clustering newlinegrouping the data into meaningful sets and validating the formed clusters newlineThe visual methods for various data analysis problems have been newlineextensively studied and the abstract data have been represented visually to newlineamplify cognition Visualization is considered to be one of the most newlineinstinctive methods for cluster detection and validation especially for newlineperforming well on the depiction of irregularly shaped clusters as a newlinepreclustering method The visual data mining allows the data miners and newlineanalysts to evaluate monitor guide the inputs products and process from the newlineresults of visualization techniques The Visual Clustering Analysis VCA is a newlinewide assortment of image processing techniques information visualization newlineand cluster analysis techniques The visualization used in the cluster analysis newlinemaps the highdimensional data with a 2Dimensional space and aids the newlineusers to have an intuitive and easily understood graph or image to newline newline
Pagination: xx, 184p
Appears in Departments:Faculty of Science and Humanities

Files in This Item:
File Description SizeFormat 
01_title.pdfAttached File185.84 kBAdobe PDFView/Open
02_certificate.pdf5.1 MBAdobe PDFView/Open
03_abstract.pdf92.46 kBAdobe PDFView/Open
04_acknowledgement.pdf60.43 kBAdobe PDFView/Open
05_contents.pdf154.44 kBAdobe PDFView/Open
06_chapter 1.pdf746.2 kBAdobe PDFView/Open
07_chapter 2.pdf291.68 kBAdobe PDFView/Open
08_chapter 3.pdf1.96 MBAdobe PDFView/Open
09_chapter 4.pdf2.02 MBAdobe PDFView/Open
10_chapter 5.pdf2.28 MBAdobe PDFView/Open
11_chapter 6.pdf2.23 MBAdobe PDFView/Open
12_chapter 7.pdf100.05 kBAdobe PDFView/Open
13_references.pdf114.62 kBAdobe PDFView/Open
14_publications.pdf68.15 kBAdobe PDFView/Open
15_vitae.pdf55.63 kBAdobe PDFView/Open

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