No description available.

The purpose of this study is to determine boundary conditions associated with the number of treatment groups (K), the common treatment group sample size (n), and an index of the extent to which the assumption of equality of treatment population variances is violated (Q) with regard to user confidence in application of the one-way analysis of variance F-test for determining equality of treatment population means. The study concludes that the analysis of variance F-test is robust when the number of treatment groups is less than seven and when the extreme ratio of variances is less than 1:5, but when the violation of the assumption is more severe or the number of treatment groups is seven or more, serious discrepancies between actual and nominal significance levels occur. It was also concluded that for seven treatment groups confidence in the application of the analysis of variance should be limited to the values of Q and n so that n is greater than or equal to 10 In (1/2)Q. For nine treatment groups, it was concluded that confidence be limited to those values of Q and n so that n is greater than or equal to (-2/3) + 12 ln (1/2)Q. No definitive boundary could be developed for analyses with five treatment groups.

No description available.

No description available.

No description available.

Discriminant analysis uses labelled observations to infer the labels of unlabelled observations in a population. Despite many advances in unsupervised and, to a lesser extent, semi-supervised learning over the past decade, discriminant analysis is often employed using approaches that date back to very well-known work of Fisher in the 1930s. One notable exception is mixture discriminant analysis, where the labels are estimated using parametric finite mixture models, commonly the Gaussian mixture model. The supposed advantage with mixture discriminant analysis is that multiple Gaussian components can be used for each class, hence providing a work around when a class is not Gaussian. This thesis makes several contributions to ``modern" discriminant analysis. Three robust discriminant analysis methods are introduced using mixtures of multivariate t-distributions, mixtures of multivariate power exponential distributions, and mixtures of contaminated Gaussian distributions, respectively. This provides an appealing framework for handling varying tail-weights and peakedness in the classes that may also contain mild outliers. To facilitate the modelling of asymmetric classes, we also explore robust discriminant analysis via finite mixtures of generalized hyperbolic distributions and mixtures of multivariate skew-t distributions. These approaches are tailored towards skewed classes but also have the added advantage of modelling symmetric classes where necessary. Finally, we introduce an approach that combines support vector machines with mixture discriminant analysis. This approach defines class boundaries in the labelled observations and, in some sense, improves mixture discriminant analysis performance. Crucially, in all of our mixture modelling work, we consider the case where the number of components per class is one. The utility of the approaches introduced is demonstrated on simulated and real data sets. / Thesis / Doctor of Philosophy (PhD)

"December 2004." / Thesis (Ph.D.)--Chinese University of Hong Kong, 2004. / Includes bibliographical references (p. 221-233) / Electronic reproduction. Hong Kong : Chinese University of Hong Kong, [2012] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Electronic reproduction. Ann Arbor, MI : ProQuest Information and Learning Company, [200-] System requirements: Adobe Acrobat Reader. Available via World Wide Web. / Mode of access: World Wide Web. / Abstracts in English and Chinese.

This work is concerned with analysing the potential impact of direct changes to large- scale enterprise systems, and, in particular, how to minimise testing efforts on such changes. A typical enterprise system may consist of hundreds of thousands of classes and millions of methods. Thus, it is extremely costly and difficult to apply conventional testing techniques to such a system. Retesting everything after a change is very expensive, and in practice generally not necessary. Selective testing can be more effective. However, it requires a deep understanding of the target system and a lack of that understanding can lead to insufficient test coverage. Change Impact Analysis can be used to estimate the impacts of the changes to be applied, providing developers/testers with confidence in selecting necessary tests and identifying untested entities. Conventional change impact analysis approaches include static analysis, dynamic analysis or a hybrid of the two analyses. They have proved to be useful on small or medium size programs, providing users an inside view of the system within an acceptable running time. However, when it comes to large-scale enterprise systems, the sizes of the programs are orders of magnitude larger. Conventional approaches often run into resource problems such as insufficient memory and/or unacceptable running time (up to weeks). More critically, a large number of false-negatives and false-positives can be generated from those approaches.In this work, a conservative static analysis with the capability of dealing with inheritance was conducted on an enterprise system and associated changes to obtain all the potential impacts. Later an aspect-based dynamic analysis was used to instrument the system and collect a set of dynamic impacts at run-time. We are careful not to discard impacts unless we can show that they are definitely not impacted by the change. Reachability analysis examines the program to see “Whether a given path in a program representation corresponds to a possible execution path”. In other words, we employ reachability analysis to eliminate infeasible paths (i.e., miss-matched calls and returns) that are identified in the control-flow of the program. Furthermore, in the phase of alias analysis, we aim at identifying paths that are feasible but cannot be affected by the direct changes to the system, by searching a set of possible pairs of accesses that may be aliased at each program point of interest. Our contributions are, we designed a hybrid approach that combines static anal- ysis and dynamic analysis with reachability analysis and alias/pointer analysis, it can be used to (1) solve the scalability problem on large-scale systems, (2) reduce false-positives and not introduce false-negatives, (3) extract both direct and indirect changes, and (4) identify impacts even before making the changes. Using our approach, organizations can focus on a much smaller, relevant subset of the overall test suite instead of blindly doing their entire suite of tests. Also it enables testers to augment the test suite with tests applying to uncovered impacts. We include an empirical study that illustrates the savings that can be attained. / Thesis / Doctor of Philosophy (PhD)

The work described in this thesis concerns the synthesis and study of the coordination behaviour of the 6-anilino-2,2'-bipyridyl (Habipy), 6-N-methylanilino-2,2'-bipyridyl (mabipy), 6-piperidyl2,2'.- bipyridyl (pipbipy) and 2,2'-bipyridin-6-one (Hobipy) ligands. Chapter one reviews the coordination chemistry of the 2-aminopyridyl (Rap), 2-hydroxypyridyl (Hhp) and the 2-(2-pyridyl)-I,8-naphthyridine (pynp) ligands. These ligands are closely related to Habipy, mabipy, pipbipy and Hobipy in that they share a common NCN or NCO fragment. Thus the review of their coordination behaviour provides insight into the expected coordination of the Habipy, mabipy, pipbipy and Hobipy ligands. The synthesis and characteristaion of the novel Habipy, mabipy and pipbipy ligands are reported in Chapter two. X-ray crystal structure determinations of mabipy and pipbipy reveal that the geometry about the exocyclic nitrogen atom in both ligands is nearly planer, suggesting substantial overlap of the nitrogen lone pair orbital with the 1t electron system of the bipyridyl rings. In both mabipy and pipbipy the N3-C 10 bond lengths are shorter then normal N-C single bonds. In Chapter three -the synthesis and characterisation of copper(I) complexes containing mabipy, pipbipy and Habipy, and a copper(II) complex containing mabipy are reported. The copper(I) complexes have the general formula [Cu(l12-L)2r, where L= mabipy 1, pipbipy 2 or Habipy 3. The structures of complexes 1 and 2 are determined by X-ray crystallography. In complexes 1- 3 the bipyridyl fragments of mabipy, pipbipy and Habipy chelate while the exocyclic nitrogen atoms remain free. The crystal structures of 1 and 2 reveal that the exocyclic nitrogens have a planar geometry as was the case in the uncoordinated ligand. The crystal structure of [Cu(1l2-mabipY)2f+ (4) is determined by X-ray crystallography and is very similar to that of the copper(I) species. Coordination of the mabipy ligand in 4 is the same as that in 1 and the exocyclic nitrogen in 4 is also planar. The redox couple 4/1 is shown to be electrochemically reversible with EV2= 0.45 V. In Chapter four the synthesis and chararcterisation of dinuclear complexes containing the Ru2 2+, Ru/+, Mo24+and ~4+ cores are reported in which the abipy ligand bridges two metal centres. The complexes [Rulll-L)lCO)4], where L= abipy 5 or obipy 6, were synthesised by the r~action of the free ligands with [{RuiCO)102CCH3)2}n] in toluene. The structures of5 and 6 are determined by X-ray crystallography and show the ligands bridging the Ru(I) atoms in a head to tail fashion and occupy mutually cis positions about the octahedral Ru(I) atoms. The Ru-Ru separations in 5 and 6 are 2.668(1) and 2.671(1) A respectively. The reaction of Habipy with [Rui02CCH3)4CI]n was found to afford the mixed valence species [Ruill-abipy)(02CCH3)3CI] (7), the structure of which is determined by X-ray diffraction methods. The structure of7 reveals one abipy ligand bridging the two ruthenium atoms as in 5. The Ru-Ru separation in 7 is 2.294(2) A. The reaction of [Moi02CCH3)4] with habipy in methanol affords (Moill-abipy)(02CCH3)3] (8). The structure of 8 is determined by X-ray diffraction methods and reveals one abipy ligand bridging two quadruply bonded molybdenum atoms which have a Mo-Mo separation of 2.094(2) A. The [Rhlll~abipy)(02CCH3)iH20)] (9) is formed from the reaction of [Rh i02CCH3)4] with Habipy in methanol. The structure of [Rh2(Il-abipy)(02CCH3)iNCPh)] is determined by X-ray diffraction methods and shows the abipy ligand bridging two Rh(II) atoms which are separated by 2.399(1) A. This chapter is concluded with a discussion of the possiblity of substitution of more than one acetate ligand by the abipy ligand in terms of 'hard' and' soft' acid-base theory and synthetic methods. / Thesis (M.Sc.)-University of Natal, Pietermaritzburg, 1996.

Call number: LD2668 .T4 1985 C92 / Master of Science