-
Prospective Students
-
Faculty and Staff
-
Current Students
-
Alumni and Friends
In this thesis, we hypothesize that data visualization users are subject to systematic errors, or cognitive biases, in decision-making under uncertainty. Based on research from psychology, behavioral economics, and cognitive science, we design five experiments to measure the role of anchoring bias, confirmation bias, and myopic loss aversion under different uncertain decision tasks like social media event detection, misinformation identification, and financial portfolio allocation. This thesis makes three major contributions. First, we find evidence of cognitive biases in data visualization through multiple behavioral trace data including user decisions, interactions logs (hovers, clicks), qualitative feedback, and belief elicitation techniques. Second, we design five digital experiments with interactive data visualization systems across different design complexities (coordinated multiple views to single plot) and data types (social network, linguistic, geospatial, temporal, statistical) and evaluate them on user populations that range from novice to expert (crowdsourced, undergraduate, data scientist, domain expert). Third, we evaluate the experiments using statistical, probabilistic, and machine learning techniques to measure the effects of cognitive biases with mixed effects modeling, hierarchical clustering, natural language processing, and Bayesian cognitive modeling. These experiments show the promising role data visualizations and human-computer techniques could remediate such biases and lead to better decision-making under uncertainty.
Learning Analytics (LA) has had a growing interest by academics, researchers, and administrators motivated by the use of data to identify and intervene with students at risk of underperformance or discontinuation. Typically, faculty leadership and advisors use data sources hosted on different institutional databases to advise their students for better performance in their academic life. Although academic advising has been critical for the learning process and the success of students, it is one of the most overlooked aspects of academic support systems. Most LA systems provide technical support to academic advisors with descriptive statistics and aggregate analytics about students' groups. Therefore, one of the demanding tasks in academic support systems is facilitating the advisors' awareness and sensemaking of students at the individual level. This enables them to make rational, informed decisions and advise their students. To facilitate the advisors' sensemaking of individual students, large volumes of student data need to be presented effectively and efficiently.
Effective presentation of data and analytic results for sensemaking and decision-making has been a major issue when dealing with large volumes of data in LA. Typically, the students' data is presented in dashboard interfaces using various kinds of visualizations like scientific charts and graphs. From a human-centered computing perspective, the user’s interpretation of such visualizations is a critical challenge to design for, with empirical evidence already showing that ‘usable’ visualizations are not necessarily effective and efficient from a learning perspective. Since an advisor's interpretation of the visualized data is fundamentally the construction of a narrative about student progress, this dissertation draws on the growing body of work in LA sensemaking, data storytelling, creative storytelling, and explainable artificial intelligence as the inspiration for the development of FIRST, Finding Interesting stoRies about STudents, that supports advisors in understanding the context of each student when making recommendations in an advising session. FIRST is an intelligible interactive interface built to promote the advisors' sensemaking of students' data at the individual level. It combines interactive storytelling and aggregate analytics of student data. It presents the student's data through natural language stories that are automatically generated and updated in coordination with the results of the aggregate analytics. In contrast to many LA systems designed to support student awareness of their performance or to support teachers in understanding the students' performance in their courses, FIRST is designed to support advisors and higher education leadership in making sense of students' success and risk in their degree programs. The approach to interactive sensemaking has five main stages: (i) Student temporal data Model, (ii) Domain experts’ questions and queries, (iii) Student data reasoning, (iv) Student storytelling model, and (v) Domain experts’ reflection. The student storytelling stage is the main component of the sensemaking model and it composes four tasks: (i) Data sources, (ii) Story synthesis, (iii) Story analysis, and (iv) User interaction.
The contributions of this study are: i) A novel student storytelling model to facilitate the sensemaking of complex, diverse, and heterogeneous student data, ii) An anomaly detection model to enrich student stories with interesting, yet, insightful information for the domain experts and iii) An explainable and interpretable interactive LA model to inspire advisors' trust and confidence with the student stories. This study reports on four ethnographic studies to show the potential of the proposed LA sensemaking model and how it affects the advisor's sensemaking of students at the individual level. The user studies considered for this dissertation were focus group discussions, in-depth interviews, and diary study- in-situ and snippet technique. These studies investigate if FIRST can improve and facilitate the advisor's sensemaking of students’ success or risk by presenting individual student's heterogeneous data as a complete and comprehensive story.
While continuous improvement initiatives such as benchmarking have a history of utilization for general business objectives, their successful utilization in the built environment industries, such as construction and facilities management is not nearly as well documented or researched. This project identifies how the built environment fields are using continual improvement initiatives, evaluates how effectively these initiatives are being utilized, and identifies critical success factors for improving and leveraging these techniques to achieve the sustained continuous improvement initiatives that will be necessary to meet long -term sustainability goals in relation to the operations of the built environment. This project takes place in three parts; a case study of a novel way to benchmark and identify areas for improvement, a large-scale survey of how facility managers are using benchmarking and their involvement in benchmarking networks, and an analysis of the relationship of organizational learning culture and the role that it plays in facilitating and supporting benchmarking initiatives. This research provides the first-of its-kind survey and assessment of how practitioners in the built environment are utilizing benchmarking. The results of this project serve to assist facility practitioners in developing, leveraging, and strengthening their continuous improvement initiatives to sustain ongoing change critical for the success of long-term organizational goals related to the built environment lifecycle.
Freeform optics have bridged the gap from theoretical to practical application and is propelled by ultra-precision multi-axis machining. Freeform optics have been used for infrared sensors, vision correction, and beam shaping. Manufacturing and application of dynamic freeform optics, where relative motion of freeform surfaces can enable improved or new functionality of an optical system, is a next step. The first part of this work concentrates on evaluating various manufacturing paths for glass transmissive dynamic freeform optics. Leveraging an iterative process design and metrology techniques, a method for the generation of high-quality optics for production is established. Metrology evaluations led to development of a six degree of freedom surface analysis that utilizes simulated annealing for optimization. Major results from the precision glass molding indicate high-volume production of transmissive glass freeform optics is possible. The second part of this work details research in the manufacturing of two separate dynamic freeform optics and optomechanics. For prototyping of visibly transmissive dynamic freeforms, a shift was machined into the optical surfaces. These dynamic systems allow for novel light management and improved depth of field in high-magnification systems. All of these aforementioned freeform processes clarify the methods for future manufacturing of freeform optics and associated optomechanics.
The main goal of learning analytics and early detection systems is to extract knowledge from student data to understand students' trends of activities towards success and risk and, therefore, design intervention methods to improve learning performance and experience. However, many factors contribute to the challenge of designing and building effective learning analytics systems. Because of the complexity of heterogeneous student data, models designed to analyze it frequently neglect temporal correlations in the interest of convenience. Moreover, the performance descriptions gained from the student data model or prediction results from the analytical models do not always help explain the "why" and "how" behind it. Furthermore, domain specialists are unable to participate in the knowledge discovery process since it necessitates significant data science abilities, and an analytical model is a black box to them.
This research aims to develop analytical models that enable domain experts to study their students' performance behavior and explore trustworthy sources of information with the help of explanations on the analytics. Our work demonstrates various approaches to using the temporal aspect of heterogeneous student data to build analytical models: weighted network analysis, unsupervised cluster analysis, and recurrent neural network analytics. The description, implementation process, and findings of each method are presented as technical contributions to the temporal analysis of student data. We experiment with all these analytical models that highlight the complexity of heterogeneous-temporal data, model building, decision-making tasks, and the need for a more in-depth focus on visual information of analytics with state-of-art explainable AI tools and techniques.
Our work underscores a need for developing a robust way to integrate the possibilities inherent within each approach. To achieve this goal, we present a comprehensive yet flexible and empirical framework to support the design and development of analytical models to extract meaningful insights about students' academic performance and identify early actionable interventions to improve the learning experience. We illustrate our framework on three applications (e.g., student network model, unsupervised clustering model, and recurrent neural network analytics) to demonstrate the value of this framework in addressing the challenges of using student data for learning analytics. These applications present vast opportunities to benefit students' learning experience by implementing flexible educational data representations, fitting different predictive models, and extracting insights for designing prescriptive analytics and building strategies to overcome perceived limitations.
An academic institution's culture drives its ability to accept, leverage, and deploy predictive and prescriptive analytics to enhance the workflow of maximizing pedagogical outcomes. We believe that our work will aid in the future development or refinement of a set of design standards for learning analytics systems.
In this thesis, a combined approach based on the Finite Element (FE) and Smoothed Particle Hydrodynamics (SPH) methods is proposed to model turning operations. The approach exploits the advantages of each method and leads to high-fidelity coupled FE-SPH machining models that are significantly more numerically efficient and are on par with the models based on each of the two methods alone. Both two-dimensional and three-dimensional models are developed and validated by comparing predicted forces and chip morphologies with experimental results. Parametric studies are carried out to fine-tune the model-based parameters in order to avoid numerical stability issues. The three-dimensional models are extended to include modulated tool path (MTP) machining which is a technique for breaking chips during machining by modulating the motion of the tool. The MTP model predictions are shown to agree with the results from an existing analytical model. With this model, various tool paths can be simulated to choose an optimal path that decreases tool-wear without sacrificing productivity. Preliminary results from a three-dimensional turning model incorporating machining dynamics through a spring-damper system are also presented. This model has the potential to be used for studying machining stability for a given set of machining conditions.
In addition to the above, another significant contribution of this thesis is the determination of Johnson-Cook material model parameters for a given material using an inverse method and experimental values of cutting forces and workpiece temperatures. The methodology described in the present work identifies the non-uniqueness of the solution to the inverse problem and proposes an approach that eliminates the non-uniqueness.
Charlotte Bilingual Preschool’s Family Program aims to equip parents to support their children’s education at home and at school by increasing parents’ educational engagement, promoting parenting best practices, developing families’ social capital, and supporting families’ mental health and well-being. This study aimed to evaluate 1) how the Family Program promotes growth in these areas for the families at the preschool, 2) the interconnections among parents’ attitudes, behaviors, and supports, and 3) how parents’ attitudes, behaviors, and supports relate to their children’s functioning in preschool.
The analysis of survey data collected at the beginning and end of the 2019-20 school year uncovered little evidence that attendance at Family Program events (i.e., Family Cafes and Workshops) led to improvements in family or child outcomes, other than increased parent friendships and more connections in the preschool family network. The disruption of programming caused by the COVID-19 pandemic and the shift to remote instruction in March 2020, likely relate to the lack of findings in this area. Nonetheless, correlational and regression analyses did identify relationships among mothers’ attitudes, perceptions of social support and social capital, and educational involvement behaviors. For instance, findings suggest that common good social capital (i.e., a positive, collaborative community atmosphere) may promote positive interactions with teachers and other parents, which could in turn, promote more positive educational involvement behaviors, including home-based involvement, ethnic identity parenting, and more positive behavior management practices. Additionally, analyses indicated that the positive relationship between maternal stress and negative behavior management practices was attenuated when mothers perceive strong social support and social capital. While these positive outcomes did not relate to parents’ attendance at Family Cafes and Workshops, they were associated with parents’ self-reported school involvement, suggesting that parents’ broader interactions with the Family Program (i.e., beyond attendance at Family Cafes and Workshops) may yield positive outcomes.
This study’s findings support the approach of Charlotte Bilingual Preschool’s Family Program, by connecting caregivers’ attitudes, sense of support, and social capital (which are intermediate goals of the Family Program) to their educational involvement behaviors (the Family Program’s primary goal). Theoretically, promoting positive family involvement should yield more positive developmental outcomes for children in the short- and long-term as well. This study provided some support for this hypothesis, by connecting parents’ bonding and bridging social capital and their efforts to promote children’s appreciation of their ethnic and cultural identities to children’s social-emotional functioning and language skills. Furthermore, results suggested that when parents reported greater increases or improvements in several family-level variables, their children tended to show larger improvements in social-emotional protective factors and behavior. These findings indicate that the Family Program can have an important impact on children and families, especially by connecting socially isolated families with greater social support and social capital.
The COVID-19 pandemic and the preschool’s shift to remote programming in March 2020 was a major limitation that disrupted programming and reduced this study’s capacity to draw strong conclusions. However, the pandemic also provided an opportunity to examine the links between various forms of remote engagement and outcomes for children and families. Despite the pandemic, this study’s findings have important implications for Charlotte Bilingual Preschool, as well as other stakeholders seeking to enhance two-generation approaches to early childhood education; especially those supporting Latino immigrant families and English language learners. Limitations, implications, and future directions are discussed.
Athletes are aware that with involvement in sport they are exposed to the risk of getting injured. Suffering an injury can be one of the most stressful experiences in a student-athlete’s athletic career and can cause a series of psychological, emotional, and social responses, as well as impact one’s sense of identity. The very sparse literature in the counseling field regarding student-athletes and lack of research in general, exploring women student-athletes and women track and field student-athletes in particular, contributes to the need for this study. The purpose of this study was to explore lived experiences of former women track and field college student-athletes who trained and competed through pain and injury. This study utilized a phenomenological approach and implemented semi-structured interviews. Over the course of a six-week period, a total of 10 participants completed a demographic questionnaire and were interviewed via Zoom to facilitate in-depth descriptions of their experiences. Moustakas (1994) methods consistent with qualitative phenomenological research design were used to facilitate the data analysis. A total of five major themes emerged from the data, including: identity, perception of pain and injury, student-athlete - coach relationship, support system, and psychological impact. This research found that the themes are interconnected and impact each other. The findings indicate that women track and field student-athletes who chose to train and compete through pain and injuries face identity challenges, which are further facilitated by student-athlete – coach relationship, one’s support system, and acceptance of the “push through the pain” mindset. This mindset was found to be further facilitated by the underlying belief that student-athlete role is a job for which participants have been compensated. Participants were also found to minimize and justify their pain as a coping mechanism to help them in continuing to train and compete despite being in pain and injured. The relationship between participants and their coaches was found to contribute to negative psychological experiences. All themes were closely connected with cognitive and emotional functioning of the participants. Implications for counselors and counselor educators as well as future research recommendations are discussed. However, the emphasis for counselors is to approach working with student-athletes from a holistic standpoint, disclose personal experiences with athletics early on in the therapeutic relationship, and provide substantial psychoeducation regarding intercorrelation between mental health and athletic performance.
