Welcome to Research Commons

Research Commons is the University of Waikato's open access research repository, housing research publications and theses produced by the University's staff and students.

Communities in Research Commons

Select a community to browse its collections.

Now showing 1 - 2 of 2

Recent Submissions

Branches: Efficiently seeking optimal sparse decision trees via AO*
(PMLR, 2025) Chaouki, Ayman; Read, Jesse; Bifet, Albert
Decision Tree (DT) Learning is a fundamental problem in Interpretable Machine Learning, yet it poses a formidable optimisation challenge. Practical algorithms have recently emerged, primarily leveraging Dynamic Programming and Branch & Bound. However, most of these approaches rely on a Depth-First-Search strategy, which is inefficient when searching for DTs at high depths and requires the definition of a maximum depth hyper-parameter. Best-First-Search was also employed by other methods to circumvent these issues. The downside of this strategy is its higher memory consumption, as such, it has to be designed in a fully efficient manner that takes full advantage of the problem’s structure. We formulate the problem within an AND/OR graph search framework and we solve it with a novel AO*-type algorithm called BRANCHES. We prove both optimality and complexity guarantees for BRANCHES and we show that it is more efficient than the state of the art theoretically and on a variety of experiments. Furthermore, BRANCHES supports nonbinary features unlike the other methods, we show that this property can further induce larger gains in computational efficiency.
A conceptual and psychometric examination of climate anxiety
(The University of Waikato, 2026) Gago, Tomás; Sargisson, Rebecca J.; Milfont, Taciano L.
Climate change stands as one of the most pressing crises threatening the world today, affecting natural environments, community resources and functioning, and individual’s health. As this crisis worsens, people are increasingly reporting experiencing negative affective reactions to climate change, even when they have not themselves been directly impacted by it. These various reactions, described under the umbrella term of “climate anxiety”, have been the focus of numerous studies, but the lack of a consistent definition and theory of climate anxiety has hindered scholars’ ability to integrate and build on the knowledge gained from these studies. My aim with this thesis is to contribute to this integration effort by articulating the various contributions from other scholars into a cohesive and holistic conceptualization of climate anxiety, and to examine how prevalent and adaptive this phenomenon is. Thus, the first study in my thesis (Chapter 3) is a metaanalysis of 25 studies examining the link between climate anxiety and psychological wellbeing. In this study, we found there was a strong association between these variables, whereby higher levels of climate anxiety were associated with lower levels of wellbeing. Given the consistent association between climate anxiety and psychological illbeing, in the next study (Chapter 4), we found that measures of climate anxiety capture this phenomenon differently to how measures of domain-free psychopathology capture their respective target constructs, by focusing more on affective reactions and less on somatic complaints. Additionally, we also found evidence supporting the convergence of the different climate anxiety measures, indicating they could be capturing the same underlying trait, and raising questions about the lack of parsimony in climate anxiety measuring. To address these concerns, in Chapter 5, we report a mixed-methods study where we developed a condensed climate anxiety measure with a wider representation of the various features associated with climate anxiety using Rasch modelling. Despite its wider content representation and ability to differentiate between high levels of climate anxiety, our measure did not predict wellbeing outcomes better than an existing eco-anxiety measure. Therefore, for the next study (Chapter 6), we used an established measure of climate anxiety to assess the prevalence of climate anxiety in New Zealand, finding that 1 in 20 New Zealanders report experiencing these affective reactions. Furthermore, we found that people who have been directly impacted by climate change, people with environment-dependent jobs, and those who did not feel prepared to handle climate change reported higher levels of climate anxiety. Lastly, we found that those with higher levels of climate anxiety also reported higher psychological distress and more engagement in pro-environmental behaviours. Altogether, this thesis contributes to the climate anxiety literature by providing a framework for the assessment of the various dimensions of climate anxiety and how they may be contributing to individual and planetary wellbeing. Moreover, the findings here may be used to raise awareness for the potential benefits and impairment associated with climate anxiety, and justify support for groups at increased risk of maladaptive climate anxiety.
SGLT2 inhibitor use and disparities in all-cause mortality in type 2 diabetes: Insights from a multi-ethnic population
(Springer, 2026) Chepulis, Lynne; Gan, Han; Simmons, David; Rodrigues, Mark William; Keenan, Rawiri; Murphy, Rinki; Kenealy, Tim; Te Karu, Leanne; Magliano, Dianna; Scott-Jones, Jo; Moffitt, Allan; Lao, Chunhuan; Lawrenson, Ross; Paul, Ryan G.
Aims/hypothesis: Sodium-glucose cotransporter 2 inhibitors (SGLT2i) are known to reduce cardiovascular and all-cause mortality in people with type 2 diabetes, but there are limited data regarding mortality outcomes in different ethnic groups (including Indigenous peoples). This study reports on mortality outcomes in a population in Aotearoa New Zealand (hereafter New Zealand) with type 2 diabetes, following the funded availability of the SGLT2i empagliflozin with prioritised access for Māori and Pacific people. Methods: Data were collected from primary care records for those aged 18–75 years with type 2 diabetes (Auckland/Waikato regions of New Zealand; February 2021 to December 2023; n=59,505). These data were linked to national medication-dispensing and mortality records for 2021–2024 via national health identifier numbers. Following propensity matching and Cox modelling for ethnicity, age, gender, medication use, baseline HbA1c and cardiovascular and/or renal disease/risk (CVRD) status (yes/no), mortality rates were compared by ethnicity in those with and without CVRD and who did/did not initiate empagliflozin. This study was reported in accordance with the CONSIDER statement, used to strengthen the reporting of research involving Indigenous peoples. The study was funded by the Health Research Council of New Zealand. Results: Following matching, two groups of 12,792 individuals were identified. Annualised crude mortality (deaths per 1000 individuals per year) was higher in those not dispensed with SGLT2i than in those receiving SGLT2i (35.2 vs 13.1 in those with CVRD and 7.7 vs 3.6 in those without CVRD, respectively). After adjustment, the greatest difference in mortality with SGLT2i use was seen in Māori (HR 0.475; 95% CI 0.336, 0.672; p<0.001), followed by Pacific people (HR 0.507; 95% CI 0.395, 0.651; p<0.001) and European people (HR 0.667; 95% CI 0.545, 0.816; p<0.001). Conclusions/interpretation: The protective effect of SGLT2i use on mortality appears to differ by ethnicity and is greater in Indigenous Māori and Pacific populations in New Zealand with type 2 diabetes. SGLT2i use in Indigenous and minority populations may support improved health equity.
Slab turbulence in the very local interstellar medium and the IBEX ribbon
(IOP Publishing, 2025) Zirnstein, Eric J.; McComas, David J.; Giacalone, Joe; Zank, Gary P.; Guo, Fan; Heerikhuisen, Jacob; Li, Hui; Reisenfeld, Daniel B.
In this study, we analyze an important property of the very local interstellar medium, i.e., turbulence that affects the intensity and shape of the Interstellar Boundary Explorer (IBEX) ribbon. Specifically, we simulate the propagation of the ribbon’s parent ions before they become secondary energetic neutral atoms that can be observed at 1 au by IBEX. We then test how different slab/2D turbulence fractions affect the intensity and shape of the modeled ribbon and compare to the latest IBEX ribbon-separated data. We compare 1D cuts across the modeled and observed ribbon as a function of angle away from the ribbon center. We find that the intensities of the modeled ribbon are larger than most of the data, except near the ecliptic plane. However, the model intensities are sensitive to how we model the neutral solar wind, which forms the source ion population for the ribbon. We then compare the model and data by normalizing the fluxes to their respective peak intensities and find the peaks’ angular distances from the ribbon center. We find that most of the model peaks lie within 3° of the data peaks, which we consider our 1σ uncertainty (half the size of an IBEX pixel). By averaging the instances where the model peaks match the data peaks (within 3°), we find that the mean is slab-50%, with a standard deviation of ±28% and standard error of ±5%.
A novel inversion method for electrical impedance tomography with a radial basis operator network
(MDPI, 2026) Kurz, Jason A.; Pangia, Andrew; Khan, Taufiquar
We apply a new operator neural network to solve the Electrical Impedance Tomography (EIT) inverse problem. The EIT inverse problem involves reconstructing the conductivity inside a specific body or domain, given the electric potential along the boundary of said body. Mathematically speaking, the inverse problem is known to be severely ill-posed, that is, hard to reliably solve. However, we demonstrate the efficacy of our proposed algorithm utilizing the aforementioned neural network, dubbed the Radial Basis Operator Network (RBON) in its seminal work, when applied to the EIT inverse problem.