Please use this identifier to cite or link to this item: http://hdl.handle.net/10603/7197
Full metadata record
DC FieldValueLanguage
dc.coverage.spatialStatisticsen_US
dc.date.accessioned2013-02-28T10:27:54Z-
dc.date.available2013-02-28T10:27:54Z-
dc.date.issued2013-02-28-
dc.identifier.urihttp://hdl.handle.net/10603/7197-
dc.description.abstractNormal distribution is one of the most celebrated statistical distribution in the literature. The importance of normal distribution is mostly because of central limit theorem, one of the fundamental theorem that form a bridge between Probability and Statistics and because of its extensive applications in various scientific investigations. A large number of alternative distributions were suggested in place of normal probability densit function. For example, when the data seems to be more heavy tailed than normal, Laplace distribution can be used instead of normal. When the information is available in the form of first few moments we use normality assumption when the skewness ¯1 = 0 and kurtosis ¯2 = 3. Kale and Sebastian (1996) showed that there exist a wide class of symmetric distributions with Pearsons measure of kurtosis ¯2 = 3. A member of this class can be obtained by considering a mixture of two symmetric non-normal densities, with centers of symmetry being the same, say zero, the kurtosis of one component strictly less than 3 and that of the other component strictly greater than 3. Motivated from the examples of Kale and Sebastian (1996) in this thesis we try to identify and characterize a fairly large class of symmetric mesokurtic distributions and try to replace this class in place of normal model. We, then study the properties of the estimators, when the random sample is from one of the members of this family, in which normal distribution is a particular case. The thesis is divided into five chapters. Chapter 1 provides an introduction and summary of the thesis. In Chapter 2, we prove few characterization results of symmetric mesokurtic family. While all non-normal mesokurtic densities considered by Kale and Sebastian (1996) and others were mixtures of two symmetric probability density functions, we give an example of a class of mesokurtic distribution which cannot be considered as a mixture of any two densities. This class contains both unimodal as well as multimodal densities.en_US
dc.format.extent101p.en_US
dc.languageEnglishen_US
dc.relation-en_US
dc.rightsuniversityen_US
dc.titleSome properties and inferential problems related to non- normal distributions with Kurtosis Threeen_US
dc.title.alternative-en_US
dc.creator.researcherKurian, Jamesen_US
dc.subject.keywordRobustnessen_US
dc.subject.keywordEdgeworth expansionen_US
dc.subject.keywordGaussian error modelen_US
dc.subject.keywordLocation-scale familyen_US
dc.subject.keywordMixture distributionsen_US
dc.subject.keywordMML estimationen_US
dc.subject.keywordMoment measure of kurtosisen_US
dc.description.note-en_US
dc.contributor.guideGeorge, Sebastianen_US
dc.publisher.placeKottayamen_US
dc.publisher.universityMahatma Gandhi Universityen_US
dc.publisher.institutionDepartment of Statisticsen_US
dc.date.registeredn.d.en_US
dc.date.completedJanuary 2008en_US
dc.date.awardedn.d.en_US
dc.format.dimensions-en_US
dc.format.accompanyingmaterialNoneen_US
dc.type.degreePh.D.en_US
dc.source.inflibnetINFLIBNETen_US
Appears in Departments:Department of Statistics

Files in This Item:
File Description SizeFormat 
01_title.pdfAttached File25.53 kBAdobe PDFView/Open
02_declaration.pdf42.31 kBAdobe PDFView/Open
03_certificate.pdf42.3 kBAdobe PDFView/Open
04_acknowledgements.pdf56.49 kBAdobe PDFView/Open
05_abstract.pdf92.46 kBAdobe PDFView/Open
06_contents.pdf90.7 kBAdobe PDFView/Open
07_chapter 1.pdf191.61 kBAdobe PDFView/Open
08_chapter 2.pdf435.68 kBAdobe PDFView/Open
09_chapter 3.pdf454.98 kBAdobe PDFView/Open
10_chapter 4.pdf203.76 kBAdobe PDFView/Open
11_chapter 5.pdf327.08 kBAdobe PDFView/Open


Items in Shodhganga are licensed under Creative Commons Licence Attribution-NonCommercial-ShareAlike 4.0 International (CC BY-NC-SA 4.0).

Altmetric Badge: