Please use this identifier to cite or link to this item: http://hdl.handle.net/10603/22974
Title: Bagel A Novel Hybrid Technique For Missing Value Estimation In Mixed Attribute Datasets
Researcher: Devi Priya R
Guide(s): Kuppuswamy S
Keywords: Bagel
mean mode values
Missing Value Estimation
Mixed Attribute Datasets
Novel Hybrid Technique
Upload Date: 19-Aug-2014
University: Anna University
Completed Date: n.d.
Abstract: In todays modern world tremendous increase in computer usage newlinedemands storage and retrieval of large amount of data from heterogeneous newlinesources Various data mining techniques are used to discover knowledge from newlinelarge datasets In such datasets incomplete data is a very common problem newlinefaced by analysts Analysis made from such incomplete data may not be newlinealways accurate and lead to misleading decisions Hence some efficient newlinestatistical techniques need to be applied to fill the missing values newlineNumerous techniques have been proposed by researchers to handle newlinethis issue Some simple methods like substituting zero random values default newlinevalues and mean mode values for missing values are commonly followed newlineThey are very simple to use but they assume that all missing values exhibits newlinethe same behavior and replace all their positions with a single value Hence newlinethey are generally not recommended by researchers newlineSome statistical techniques like Knearest neighbor classification newlineKmeans clustering Multiple Imputation Maximum likelihood approaches newlinePrincipal Component Analysis Genetic Algorithms Neural networks newlineBayesian techniques and Kernel estimator are used But application of these newlinetechniques demands more statistical knowledge from the user Also all the newlinetechniques cannot be applied in all scenarios Depending upon the nature of newlinemissingness and type of the attribute technique needed also may vary A newlinegreat challenge faced by analysts and researchers is to search appropriate newlinetechnique suitable for their problem Simple methods are not efficient and newlineefficient methods are not simple enough for all users to understand This newlinesituation was the motivating factor behind this research to develop a novel newlinemethod that can be applied for all kinds of missing mechanisms and newlineattributes newlineThis research proposes a new approach called Bayesian Genetic newlineAlgorithm BAGEL which combines the features of Bayesian principles and newlineGenetic Algorithms to handle missing data problem newline
Pagination: xxvi, 201p
URI: http://hdl.handle.net/10603/22974
Appears in Departments:Faculty of Information and Communication Engineering

Files in This Item:
File Description SizeFormat 
01_title.pdfAttached File25.26 kBAdobe PDFView/Open
02_certificate.pdf962.43 kBAdobe PDFView/Open
03_abstract.pdf82.13 kBAdobe PDFView/Open
04_acknowledgement.pdf8.49 kBAdobe PDFView/Open
05_content.pdf64.28 kBAdobe PDFView/Open
06_chapter 1.pdf65.62 kBAdobe PDFView/Open
07_chapter 2.pdf236.84 kBAdobe PDFView/Open
08_chapter 3.pdf23.35 kBAdobe PDFView/Open
09_chapter 4.pdf146.53 kBAdobe PDFView/Open
10_chapter 5.pdf624.76 kBAdobe PDFView/Open
11_chapter 6.pdf154.41 kBAdobe PDFView/Open
12_chapter 7.pdf266.82 kBAdobe PDFView/Open
13_chapter 8.pdf59.43 kBAdobe PDFView/Open
14_chapter 9.pdf15.17 kBAdobe PDFView/Open
15_reference.pdf808.57 kBAdobe PDFView/Open
16_publications.pdf26.1 kBAdobe PDFView/Open
17_vitae.pdf7.36 kBAdobe PDFView/Open


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