82837 research outputs found
Sort by
Partition-Free Cluster Evaluation: Extending Cluster Validation and Cluster Extraction from Hierarchies
No full textRunning clustering algorithms with multiple parameters reveals structures at varying levels of granularity. In density-based hierarchical clustering, these structures appear across different density levels and smoothing parameters. Cluster validation seeks to assess the quality of clustering solutions by assigning a numerical score based on solution quality. However, existing internal cluster validation indices evaluate solutions as whole partitions, limiting their ability to combine clusters across varying density levels and granularities.
We propose the Partition-Free Cluster Evaluation (PFCE) index, a density-based cluster validation index that evaluates clusters individually, enabling flexible extraction from hierarchical clustering solutions. PFCE leverages relative density estimates of separation and cohesion to assess cluster quality. Additionally, we introduce a new cluster quality factor, Dividededness, which measures the homogeneity of a cluster’s internal structure, making PFCE responsive to subcluster structures within clusters.
Our approach integrates PFCE within the HDBSCAN* framework to efficiently extract clusters from hierarchies while utilizing HDBSCAN* density estimates. Unlike traditional optimization criteria for extracting partitions, PFCE's cluster qualities are directly comparable across clusters with different density estimates, offering a unified and efficient method for both cluster extraction and inter-partition comparison.
Through empirical validation on real-world, synthetic, and benchmark datasets, we demonstrate that PFCE reliably evaluates partitions containing noise and effectively handles clusters with varying geometries, densities, and inter-cluster separations. We also find that PFCE outperforms existing extraction methods, showing that it can extract and compare clusters from multiple HDBSCAN* hierarchies. By reformulating cluster validation on a per-cluster basis, this research advances toward more flexible clustering methodologies, enabling the integration of clusters from multiple clustering solutions, including different parameters and clustering algorithms
Game Meat Motivation Among Hunters in Alberta
No full textHunting plays a vital role in North American culture - supporting rural economies and assisting wildlife management, making the activity a fundamental part of wildlife conservation. While it is no longer essential for survival, many still engage in and advocate for hunting. However, the views of hunting and hunters are changing within Canadian societies. Divergent perceptions of the activity influence hunters who feel increasingly ambiguous about their participation in hunting. To encourage greater participation, particularly among casual hunters and even more so among active hunters, a potential strategy would be to emphasize the value of hunting through the game meat it provides. A survey was conducted to identify perceptions and explore the food-related motivations influencing game meat consumption among hunters in Alberta by employing the Free Word Association Technique (FWAT) and Best-Worst Scaling (BWS) experiment. The findings indicate that the primary motivations for hunting are to spend time outdoors and acquire meat for food. Moreover, the study revealed that both active and casual hunters primarily base their perceptions of game meat on health and food-related attributes, with hunting-related views being of less importance. In exploring the food-related motives of hunters, the study established that hunters prioritize qualities like freshness, taste, self-hunting, and safety over attributes such as calorie content, fat content, and ease of obtaining and preparation. Notably, casual hunters showed a stronger preference for quality and freshness, appealing taste, safety, and game meat they had hunted themselves. They also emphasized the importance of easy preparation and accessibility of the meat. This study provides an in-depth analysis of hunters’ values, perceptions, and preferences for game meat attributes, presenting insights that enrich our understanding of modern hunting culture which can guide and offer practical information to wildlife conservation agencies and stakeholders
Guide to Perennial Horticultural Plants suitable for 2024 Canadian Plant Hardiness Zone 4
No full textRobust Design and Misspecification Detection for Count Models
No full textCount models, such as the Poisson and negative binomial models, are widely utilized in various fields, including epidemiology, engineering, and social sciences, to model the occurrence of events within fixed periods of time or space. Despite their extensive application, these models are highly susceptible to misspecification, which occurs when the assumed form of the model deviates from the true underlying process. Misspecification can result in biased estimates and invalid inferences, emphasizing the need for robust design methodologies. This thesis addresses the gap in robust design strategies tailored specifically for count models by developing and evaluating two robust design methodologies aimed at enhancing model robustness against misspecification and detecting misspecification in count models. Leveraging existing robust design methodologies, robust optimization techniques, and model diagnostics, this work aims to improve the reliability and validity of statistical inferences drawn from count data. The thesis is structured as follows: a literature review on robust design methodologies and their applications, theoretical frameworks for count models, optimal design, and misspecification detection, development of loss functions and optimization techniques using simulated annealing, experimental evaluations, and a conclusion with suggestions for future research. The proposed methodologies aim to advance the robustness of count models, providing researchers with tools to ensure model validity and account for misspecification
Comparative Analysis of Curriculum Design and Teaching Methods in Environmental Design: A Focus on China and Canada
No full textThis capping exercise presents a comparative analysis of environmental design programs in higher education institutions in China and Canada. While both countries offer undergraduate degrees in this field, the disciplinary positioning, curricular structure, and pedagogical approaches differ significantly. In China, environmental design is generally situated within the broader category of art and design education, emphasizing cultural aesthetics and spatial expression. In contrast, Canadian programs are often located within either art and design institutions (e.g., University of Manitoba) or architecture faculties (e.g., University of Calgary), focusing more on sustainability, professional skills, and interdisciplinary integration. Drawing on institutional documents, curriculum frameworks, and scholarly literature, this study examines how socio-cultural, policy, and educational contexts shape the formation and delivery of environmental design education. The paper further explores the impact of neoliberal educational reforms and global market pressures on the creative disciplines, highlighting tensions between artistic individuality and standardized, employability-driven outcomes. The findings provide insight into how the integration of theory, practice, and cultural relevance can inform future curriculum development in environmental design education in both national and global contexts
Improving Sample Integrity in Human Fecal Metabolomics: A Comprehensive Study of Storage, Homogenization, and Stability Solutions with Applications to Crab Symbiosis
No full textHuman health research is advancing constantly towards the development of new diagnostic tools and improving the quality of life. Gastrointestinal (GI) disorders are among the most prevalent health challenges, and there is growing recognition of the gut's influence on overall health, including the realization that gut microbes play an important role in “training” the immune system. Indeed, the gut microbiome is at the heart of many gut diseases and may hold the key to understanding the interactions between the microbiota and host health. Despite its importance, diagnostic methods for GI conditions or assessing gut health primarily rely on invasive procedures, like endoscopies and colonoscopies. These require significant patient preparation, which can be unpleasant and time consuming.
Feces have come to the forefront of metabolomics studies, offering insights into gut and gut microbiome health through non-invasive analysis. However, the transition of fecal metabolomics into clinical applications faces major barriers, particularly in the standardization of sample handling, storage, and preparation. Human feces contain undigested food, diverse microbiota, and active digestive enzymes. Such a complex matrix presents challenges for metabolomic analysis, as ongoing enzymatic and microbial activity can alter metabolite profiles after collection. Consequently, halting these activities immediately upon collection is critical, yet standardized protocols for achieving this are lacking.
Ensuring minimal changes to the fecal metabolite profile post-collection currently is addressed by flash-freezing or lyophilization. However, studies have not compared these methods to freshly collected feces. Moreover, fecal heterogeneity complicates reproducibility, as inadequate homogenization can lead to inconsistent data. Bead-beating homogenization, widely used in genomics and transcriptomics, offers customizable options for achieving adequate sample homogenization while enabling intracellular metabolite extraction. However, its application in fecal metabolomics remains underexplored.
Ambient temperature metabolite stabilization of fecal samples also poses a challenge. Current methods often require immediate on-site processing and storage of fecal samples, reducing patient compliance and limiting widespread adoption of fecal metabolomics studies. Although fecal stabilization kits are available, their focus on preserving microbial DNA and RNA renders them ineffective for metabolite stabilization, with some chemical preservatives being incompatible with sensitive analytical instruments. DNA Genotek’s OMNImet®•GUT device was introduced to address this limitation; however, its efficacy has not been compared thoroughly with other chemical or ambient temperature metabolite preservation methods.
This thesis addresses these knowledge gaps by conducting an in-depth evaluation of fecal sample storage and preparation methods using comprehensive two-dimensional gas chromatography coupled to time-of-flight mass spectrometry (GCGC-TOFMS), which is noted for its enhanced separation power and its ability to identify metabolites (including volatile metabolites) in complex biological samples, such as feces. The work compares fresh, frozen (–80 °C), and lyophilized feces, examining their effects on volatile/semi-volatile and non-volatile polar metabolites. It also investigates bead-beating with varying bead sizes and mixtures to optimize fecal sample homogenization. Additionally, six metabolite stabilization solvents, including the OMNImet®•GUT chemistry (ME-200), were assessed for their efficacy in preserving metabolites at ambient temperature over seven days.
The key findings from this work include: (1) lyophilization significantly enhanced metabolite recovery compared to fresh and frozen samples, with volatile metabolites demonstrating heightened sensitivity to storage conditions; (2) bead-beating with a combination of bead sizes and organic solvent achieved optimal homogenization, improving reproducibility and metabolite recovery without requiring lyophilization; and (3) stabilization solvents minimized metabolite profile alterations in fecal samples stored at ambient temperature, with ME-200 outperforming other chemistries for global metabolomics. However, given the varying extraction efficiencies of different solvents, a compound-class-based approach is recommended for solvent selection in targeted metabolomics studies. The findings outlined in this thesis aim to guide the development of standardized fecal handling methods, enhancing the reliability of fecal metabolomics and enabling its integration into clinical testing and general health research.
Finally, these innovations in fecal sample preparation hold promise for handling and analyzing other complex biological samples. To demonstrate this, the same methods developed here were applied successfully to crab gill samples, ensuring effective homogenization and metabolite extraction. This facilitated the study of symbiotic interactions between crabs and their gill microbiota through the identification of secondary metabolites, such as terpenoids, vitamin D derivatives, cycloserine, trigonelline, and amphetamine. These results showcase the broader applicability of these techniques for other non-liquid biosamples
Trace elements in peat bog waters: distinguishing natural from anthropogenic atmospheric inputs, and their potential bioavailability
No full textDust generating from open-pit mining and associated activities can influence the surrounding environment by releasing trace elements (TEs) through the dissolution of particles in both aquatic and terrestrial ecosystems. Ombrogenic peat bogs, which rely exclusively on atmospheric deposition and have a low pH (< 4) along with abundant organic acids in their waters, provide an ideal environment for the dissolution of chemically reactive mineral phases, consequently releasing TEs associated with them. In this study, TEs were analyzed in waters from living Sphagnum moss layers and peat porewaters collected from different bog sites that range in distance toward industry to evaluate the reactivity of dust. The impact of dust from mining activities on the release of TEs to bog waters was explored relative to background TE concentrations in bog waters from remote areas. Furthermore, the association of TEs with colloidal species in the dissolved fraction (<0.45 µm) was investigated to differentiate and quantify the truly dissolved species (ionic and small molecular species <0.5 kDa) to evaluate the potential bioavailability of TEs in bog ecosystems experiencing substantial dust deposition.
In both surface moss and underlying peat porewaters, most TEs, particularly lithophile elements such as Al, Li, Sr, Th, and Y, increased in concentration toward the mining area. These trends in dissolved TEs in peat bog waters align with the patterns of dust deposition rates calculated from the acid-insoluble ash content of Sphagnum mosses and peat cores from previous studies, providing evidence of dust dissolution. Notably, the higher concentrations of these lithophile elements in moss waters compared to peat porewaters are linked likely to the higher rate of contemporary dust deposition compared to the past.
In moss waters, the predominant influence of dust deposition in the mining area was observed on the dissolved concentrations of lithophile elements, such as Fe, Y, and Th, relative to remote locations. These elements were enriched >100x close to the mining area compared to the reference sites. In contrast, bitumen-enriched (Ni and V) and potentially toxic elements (e.g., As, Pb, and Sb) were less influenced and showed enrichments only up to 40 and 20x, respectively. On the other hand, in peat porewaters the enrichments of all elements were <10x, except for Mn, at sites in the vicinity of mining activities relative to the reference site. This indicates the modest impact of the reactivity of dust deposited in the past.
In both moss waters and peat porewaters from sites in the ABS region, from 10% to over 90% of most dissolved TEs were associated with colloids. In moss waters, lithophile elements (e.g., Al, Ba, Sr, Th, and Y) were primarily associated with inorganic colloids. In contrast, in peat porewaters, they were either associated with organic colloids (e.g., Al, Th, and Y) or predominantly present in the truly dissolved fraction (e.g., Sr and Ba). This variation with depth could be attributed to factors such as longer reaction times in deeper peat porewaters compared to surface moss waters, changing sources of dust due to industrial development in the area, variations in their mineralogical composition, and/or variations in the physico-chemical properties (e.g., pH, dissolved organic matter (DOM), dissolved oxygen, and redox potential) of both moss and peat porewaters.
Overall, these findings highlight the potential of using both moss and peat porewaters to monitor the reactivity of atmospheric dust inputs. Despite significantly higher rates of dust deposition closer to open-pit mining activities, the dissolved concentrations of chalcophile elements were enriched only up to 20x relative to remote locations. Further, the association of most TEs with colloids, suggesting lower bioavailability than predicted by dissolved concentrations (i.e. <0.45 µm) alone. This highlights the importance of assessing the chemical reactivity and TEs release in bioavailable forms to accurately evaluate potential environmental impact. Rather than solely measuring total dust and TE deposition, understanding their dissolution dynamics is critical for assessing ecosystem health and informing mitigation strategies