Since new product designs have little field data available a correlation between field and accelerated test life cannot be made. However, a step partially accelerated life test approach where samples are tested under normal conditions for a time and then run to failure on an accelerated test can be used to estimate the statistical model parameters. This thesis developed the maximum likelihood parameter estimates for a step partially accelerated life test based on a Weibull distribution model for a hypothetical automotive component. Using a Monte Carlo approach with type-II censoring, the effect of sample size and length of sampling period used on the variability of the estimated parameters was examined. A smaller sampling period and small sizes lead to significant variability, which decreased as the sampling period and sample size increased. Use of a partitioned sample did not lead to an improvement in the variability of the estimates. / Department of Mathematical Sciences

For reasoning under uncertainty the Bayesian network has become the representation of choice. However, except where models are considered 'simple' the task of construction and inference are provably NP-hard. For modelling larger 'real' world problems this computational complexity has been addressed by methods that approximate the model. The Naive Bayes classifier, which has strong assumptions of independence among features, is a common approach, whilst the class of trees is another less extreme example. In this thesis we propose the use of an information theory based technique as a mechanism for inference in Singly Connected Networks. We call this a Mutual Information Measure classifier, as it corresponds to the restricted class of trees built from mutual information. We show that the new approach provides for both an efficient and localised method of classification, with performance accuracies comparable with the less restricted general Bayesian networks. To improve the performance of the classifier, we additionally investigate the possibility of expanding the class Markov blanket by use of a Wrapper approach and further show that the performance can be improved by focusing on the class Markov blanket and that the improvement is not at the expense of increased complexity. Finally, the two methods are applied to the task of diagnosing the 'real' world medical domain, Acute Abdominal Pain. Known to be both a different and challenging domain to classify, the objective was to investigate the optiniality claims, in respect of the Naive Bayes classifier, that some researchers have argued, for classifying in this domain. Despite some loss of representation capabilities we show that the Mutual Information Measure classifier can be effectively applied to the domain and also provides a recognisable qualitative structure without violating 'real' world assertions. In respect of its 'selective' variant we further show that the improvement achieves a comparable predictive accuracy to the Naive Bayes classifier and that the Naive Bayes classifier's 'overall' performance is largely due the contribution of the majority group Non-Specific Abdominal Pain, a group of exclusion.

No description available.

Research Doctorate - Doctor of Philosophy (PhD) / The rapid pace of growth in the field of human genetics has left researchers with many new challenges in the area of security and privacy. To encourage participation and foster trust towards research, it is important to ensure that genetic databases are adequately protected. This task is a particularly challenging one for statistical agencies due to the high prevalence of categorical data contained within statistical genetic databases. The absence of natural ordering makes the application of traditional Statistical Disclosure Control (SDC) methods less straightforward, which is why we have proposed a new noise addition technique for categorical values. The main contributions of the thesis are as follows. We provide a comprehensive analysis of the trust relationships that occur between the different stakeholders in a genetic data warehouse system. We also provide a quantifiable model of trust that allows the database manager to granulate the level of protection based on the amount of trust that exists between the stakeholders. To the best of our knowledge, this is the first time that trust has been applied in the SDC context. We propose a privacy protection framework for genetic databases which is designed to deal with the fact that genetic data warehouses typically contain a high proportion of categorical data. The framework includes the use of a clustering technique which allows for the easier application of traditional noise addition techniques for categorical values. Another important contribution of this thesis is a new similarity measure for categorical values, which aims to capture not only the direct similarity between values, but also some sense of transitive similarity. This novel measure also has possible applications in providing a way of ordering categorical values, so that more traditional SDC methods can be more easily applied to them. Our analysis of experimental results also points to a numerical attribute phenomenon, whereby we typically have high similarity between numerical values that are close together, and where the similarity decreases as the absolute value of the difference between numerical values increases. However, some numerical attributes appear to not behave in a strictly `numerical' way. That is, values which are close together numerically do not always appear very similar. We also provide a novel noise addition technique for categorical values, which employs our similarity measure to partition the values in the data set. Our method - VICUS - then perturbs the original microdata file so that each value is more likely to be changed to another value in the same partition than one from a different partition. The technique helps to ensure that the perturbed microdata file retains data quality while also preserving the privacy of individual records.

Thesis (Ph. D.)--University of Washington, 2007. / Vita. Includes bibliographical references (p. 200-214).

Diss. Uppsala : Uppsala universitet, 2007.

Thesis (Ph. D.)--University of Missouri-Columbia, 1997. / Typescript. Vita. Includes bibliographical references (leaves 96-98). Also available on the Internet.

Thesis (Ph. D.)--University of Missouri-Columbia, 1997. / Typescript. Vita. Includes bibliographical references (leaves 96-98). Also available on the Internet.

Thesis (Ph. D.)--University of Missouri-Columbia, 2002. / Typescript. Vita. Includes bibliographical references (leaves 154-170). Also available on the Internet.

Thesis (Ph. D.)--University of Missouri-Columbia, 2002. / Typescript. Vita. Includes bibliographical references (leaves 154-170). Also available on the Internet.