School of Mathematics and Statistics - Theses

Permanent URI for this collection

Search Results

Now showing 1 - 9 of 9
  • Item
    Thumbnail Image
    Quantitative Epidemiology: A Bayesian Perspective
    Zarebski, Alexander Eugene ( 2019)
    Influenza inflicts a substantial burden on society but accurate and timely forecasts of seasonal epidemics can help mitigate this burden by informing interventions to reduce transmission. Recently, both statistical (correlative) and mechanistic (causal) models have been used to forecast epidemics. However, since mechanistic models are based on the causal process underlying the epidemic they are poised to be more useful in the design of intervention strategies. This study investigate approaches to improve epidemic forecasting using mechanistic models. In particular, it reports on efforts to improve a forecasting system targeting seasonal influenza epidemics in major cities across Australia. To improve the forecasting system we first needed a way to benchmark its performance. We investigate model selection in the context of forecasting, deriving a novel method which extends the notion of Bayes factors to a predictive setting. Applying this methodology we found that accounting for seasonal variation in absolute humidity improves forecasts of seasonal influenza in Melbourne, Australia. This result holds even when accounting for the uncertainty in predicting seasonal variation in absolute humidity. Our initial attempts to forecast influenza transmission with mechanistic models were hampered by high levels of uncertainty in forecasts produced early in the season. While substantial uncertainty seems inextricable from long-term prediction, it seemed plausible that historical data could assist in reducing this uncertainty. We define a class of prior distributions which simplify the process of incorporating existing knowledge into an analysis, and in doing so offer a refined interpretation of the prior distribution. As an example we used historical time series of influenza epidemics to reduce initial uncertainty in forecasts for Sydney, Australia. We explore potential pitfalls that may be encountered when using this class of prior distribution. Deviating from the theme of forecasting, we consider the use of branching processes to model early transmission in an epidemic. An inhomogeneous branching process is derived which allows the study of transmission dynamics early in an epidemic. A generation dependent offspring distribution allows for the branching process to have sub-exponential growth on average. The multi-scale nature of a branching process allows us to utilise both time series of incidence and infection networks. This methodology is applied to data collected during the 2014–2016 Ebola epidemic in West-Africa leading to the inference that transmission grew sub-exponentially in Guinea, Liberia and Sierra Leone. Throughout this thesis, we demonstrate the utility of mechanistic models in epidemiology and how a Bayesian approach to statistical inference is complementary to this.
  • Item
    Thumbnail Image
    Promoting change in teacher practice through supported differentiation of instruction in mathematics
    Dermody, Bryce Gilchrist ( 2019)
    Differentiated instruction has been shown to be effective in improving student learning outcomes; however, the resulting work load can be difficult for teachers to manage. A teaching package known as the NRP (Number Resource Package) was created to support teachers to differentiate their instruction, and used effectively in two classrooms. The package allows teachers to identify their students’ current understanding using a diagnostic test and a Guttman Chart, and then provides appropriate material for the area in which students need further consolidation. It assists teachers to identify, and provide instruction for, several different knowledge levels within the one classroom. Use of the NRP in the two experimental classes was compared with five classes that did not use the NRP and continued to follow their school’s mathematics curriculum. This study involved a quasi-experimental approach, using qualitative and quantitative data. Involved were an experimental group (two teachers) and a control group (five teachers) and a total of 147 year 7 students. The research took place in a large school in western Melbourne, Australia. The qualitative data consisted of three surveys and provided information on the effectiveness of the components in the NRP. The quantitative data consisted of a pre- and a post-test completed by students in both the experimental and control groups. These tests were completed at the beginning and the end of a nine-week teaching cycle and the learning gains were determined for each student (i.e. the difference between the pre- and post-test). There was a statistically significant difference between the experimental group and the control group when these learning gains were analysed. The results demonstrated that students in the experimental group who were taught using the NRP showed greater improvement on the post-test when compared to students in the control group. It was noted that those students who performed ‘below’ the expected level and those students who performed ‘above’ the expected level showed the most improvement in the experimental group, when compared with the control group.
  • Item
    Thumbnail Image
    Cores of vertex-transitive graphs
    Rotheram, Ricky ( 2013)
    The core of a graph $\Gamma$ is the smallest graph $\Gamma^\ast$ for which there exist graph homomorphisms $\Gamma\rightarrow\Gamma^\ast$ and $\Gamma^\ast\rightarrow\Gamma$. Thus cores are fundamental to our understanding of general graph homomorphisms. It is known that for a vertex-transitive graph $\Gamma$, $\Gamma^\ast$ is vertex-transitive, and that $\left|V(\Gamma^\ast)\right|$ divides $\left|V(\Gamma)\right|$. The purpose of this thesis is to determine the cores of various families of vertex-transitive and symmetric graphs. We focus primarily on finding the cores of imprimitive symmetric graphs of order $pq$, where $p
  • Item
    Thumbnail Image
    Adolescent perceptions of the concept of randomness
    Toohey, Phillip Gerard ( 1994)
    An investigation into adolescents perceptions of concepts of randomness, with a questionnaire trialled on 75 adolescent boys between Year 7 and Year 11 in Catholic schools in Melbourne, Australia.
  • Item
    Thumbnail Image
    Modelling hydrodynamic interactions between deformable droplets
    MANICA, ROGERIO ( 2007-06)
    Understanding deformations during interaction of colloidal or nano droplets. has important implications in a wide range of applications such as flotation collection and emulsion stability. The present work is important and necessary because current models are unable to properly describe recent experimental results obtained using the Atomic Force Microscope (AFM) and Surface Force Apparatus (SFA). We revisit and improve on theories involving interacting deformable droplets. A detailed theoretical model of the experiments, which accounts for surface forces, hydrodynamic interactions, droplet deformation and AFM cantilever detection, has been developed. The new feature of the model is the use of matched asymptotic expansions to derive a new boundary condition at constant volume to obtain results that are independent of the size of the computational domain. The AFM experiments provide measurements of dynamical forces between two interacting droplets as well as between a drop and a colloidal particle of micrometre radii in solution when they are driven together and then are retracted from each other. The SFA experiments measure (time dependent) surface profiles of a millimetre size mercury drop interacting with a mica surface. Different configurations have been studied including mechanical and electrical perturbations to a thin film stabilized by double layer repulsion. Interesting experimental features were observed when approaching or retracting the mica from the mercury drop.
  • Item
    Thumbnail Image
    Methods for assessment of probability of detection for nondestructive inspections
    Harding, Catherine Ann ( 2008)
    Assessment of the probability of detection (POD) of defects with non-destructive s inspection (NDI) is essential to integrity management of safety-critical structure, particularly in the aerospace industry. Traditionally, probability of detection is measured empirically by a POD trial, in which a number of inspections are performed under field (or simulated field) conditions on a set of specimens containing representative defects. These POD trials are very expensive and for many applications are not feasible. This thesis develops improved techniques for cost-effective measurement of POD. Methods for statistical analysis of hit/miss POD data are reviewed and the behaviour of selected methods is explored using simulations. Superior methods for applying confidence limits on POD from hit/miss data are identified, which are valid for much smaller sample sizes than traditional methods, such as that found in US military handbook, MIL-HDBK-1823 (1999). Model-assisted probability of detection assessment uses physics-based or empirical models to predict POD, which has the potential to reduce the requirement for conventional POD trials, increase the portability of POD information across related applications, and provide a basis for reliability assessment for applications where trials are not feasible. In this thesis, model-assisted POD is applied to a specific non-destructive inspection procedure developed for detection of cracks at fastener holes in the lower wing skin of Royal Australian Air Force F-111 aircraft. This application of POD modelling uses data obtained from field trials and laboratory experiments for angle-beam shear-wave ultrasonic detection of fatigue cracks at fastener holes, including human factors effects. A fully representative POD trial was not feasible for this inspevtion procedure, and so this application provides important real-world context for the development of model-assisted POD.
  • Item
    Thumbnail Image
    Mathematical modelling of layering in nature
    CAFFREY, JAMES ( 2012)
    This thesis examines convection-based models to simulate the formation of structured populations. Specifically, the thesis focuses on systems that exhibit characteristic layered patterns. Such structures are important throughout the natural world and range from the molecular layers formed on substrates to the shifting rock layers within the earth's crust. Our primary motivation here, however, stems from the formation of cortical layers within the mouse brain. The creation of layers within the cortex of the mammalian brain has been of interest to researchers in developmental biology. Experiments on mice have found a counter-intuitive layering sequence. The layers are formed in an inside-out arrangement, with the deepest layers formed first and the outermost last. It is believed that the glycoprotein reelin is linked to the mechanisms behind layering. In the reeler mutant, where this glycoprotein is absent, it is observed that the layer arrangement is reversed. This has led to a number of mechanisms being proposed for how reelin may affect the neurons that form the cortical layers. No conclusive theory of this phenomenon has been established so far. Mathematical modelling provides an elegant method to evaluate the feasibility of different layering mechanisms. In this context, the thesis provides models for the purpose of understanding layer formation in nature. The replication of both age-uniform and inverted structures is a key goal. We examine the effect that modelling perspective, population models versus the individual agent models, has on our results. Results from each perspective are utilised to strengthen and augment the work. Furthermore, the major issue of volume constraints is also addressed. Modelling is used to investigate the connections between the motion of invading agents and their conversion to settled states. Base models will be initially formulated from a quite general foundation. Published experimental results are utilised as inspiration for later extensions to these base models. This research culminates with a proposed theoretical framework for cortical layer formation.
  • Item
    Thumbnail Image
    Duality groups
    AISBETT, NATALIE ( 2011)
    This thesis focuses mainly on the subclass of complex reflection groups called duality groups. We have worked towards a new classification of the duality groups. As well as this, an analysis of affine diagrams has been done to set the framework for the possibility of defining a new class of groups, affine duality groups defined analogously to affine Weyl groups. We have studied one family of affine duality groups.
  • Item
    Thumbnail Image
    Assessing metacognition
    Wilson, Jeni ( 2000)
    Researchers, educators and curriculum documents promote the importance of metacognition for student learning but much confusion in the field continues to exist about what the term 'metacognition' means. This lack of clarity creates obstacles for researchers and educators. It is difficult to teach and assess what has not been clearly defined. Because of the importance attached to assessed curriculum, a likely implication is that metacognition will not be widely embraced as a worthwhile part of the curriculum unless metacognition is clearly defined and is included as part of assessment practices. This thesis investigated the assessment of metacognition within the curriculum domain of mathematics. The study involved Year 6 students from three different schools. Conventional techniques used for monitoring metacognition are criticised in terms of their validity and reliability. The need for practical assessment tasks was identified to minimise limitations of individual techniques and attend to questions of rigor. A new multi-method approach was developed and trialled for the assessment of three key metacognitive functions: Awareness, Evaluation and Regulation. The main features of this approach were a hands-on card sorting task and a video replay used within the context of a problem based clinical interview. This thesis sets out the consequences of implementing a new method for the assessment of meta cognition. The data generated provides a detailed endorsement of a theoretical model for metacognition developed in the course of this thesis. The findings of the study call into question previous research into metacognition (both methodologically and in terms of actual findings) and shed significant light on the nature of metacognition and its use by Year 6 students in the solution of mathematical problems. The results consistently show that student metacognitive behaviour is predictable regardless of school, class or task. The study also provides a technique for assessing and researching metacognition that could be adapted for other purposes and in other contexts.