Show simple item record

dc.contributor.advisorKibua, T. K.
dc.contributor.authorOmbui, Thomas Mageto
dc.date.accessioned2012-04-20T08:24:53Z
dc.date.available2012-04-20T08:24:53Z
dc.date.issued2012-04-20
dc.identifier.urihttp://ir-library.ku.ac.ke/handle/123456789/4152
dc.descriptionThe QA 276.18.O4en_US
dc.description.abstractWe consider analysis and fitting models of the regression data in the fields, which exhibit non-constant variances, which are referred to as regression models. We focus on various approaches and procedures used in estimating the variances in such models as a way of estimating the regression parameters. Chapter 1 comprises the introduction to the subject. In the first part, Chapter 2 and 3 purely discusses the parametric approach, which is commonly used due to its outstanding features of simplicity in computation, compatibility with model assumptions and for its mathematical convenience. The procedures are fully formulated in chapter 2 and the empirical study using the same procedures is covered in Chapter 3. In the second part, Chapter 4 discusses non-parametric method. The central problems of interest are the choice of the smoothing methods, choice of the Kernel and bandwidth. In Chapter 4 we illustrate both parametric and non-parametric methods in a practical situation. A contrast and the conclusion has been done in the same chapter. All the computation has been done in Splus programming language. Table formats and other organization matters are comfortably done in Microsoft Office (Word) while graphics; figure representation and analysis are computer drawn in Microsoft Office (Excel).en_US
dc.description.sponsorshipKenyatta Universityen_US
dc.language.isoenen_US
dc.subjectRegression -- Analysis//Mathematics -- Statisticsen_US
dc.titleResiduals influence and weighting in estimation of regression parametersen_US
dc.typeThesisen_US


Files in this item

Thumbnail

This item appears in the following Collection(s)

Show simple item record