22505 research outputs found
Sort by
Event-Driven Traffic Management
No full textTraffic congestion during peak hours and large public events is a persistent challenge in urban areas, affecting mobility, economic productivity, and quality of life. While many cities are moving towards smart, data-driven traffic management, the practical effectiveness of predictive models for event-driven traffic control remains uncertain. This thesis presents an offline, data-driven feasibility study that investigates whether ma- chine learning and time-series models can predict traffic volume patterns under different conditions, including weather and the presence of events. Using a historical traffic dataset with derived trend variables, the study applies exploratory data analysis (EDA) and two predictive approaches: ARIMA for univariate time-series forecasting and XGBoost for supervised learning based on lagged traffic volumes and contextual variables. Model performance is evaluated using standard regression and classification metrics, including Mean Absolute Error (MAE), Root Mean Square Error (RMSE), accuracy, and confusion matrices. The results show that, in this dataset, the XGBoost classifier performs only marginally better than the no-information rate, and the ARIMA model has limited forecasting accuracy during event-driven peaks. In other words, neither model predicts traffic conditions during events with sufficient accuracy for operational use. Rather than demonstrating a ready-to-deploy solution, the thesis therefore provides a critical assessment of the limitations of using simple historical data and basic model configurations for event-driven traffic management. The findings highlight the need for richer, real-time data (e.g., high-resolution event information, public transport feeds, pedestrian flows) and more advanced modeling strategies before reliable event-driven traffic control can be implemented. The study contributes by: (i) systematically documenting the gap between theoretical expectations and empirical performance of common models on a realistic dataset, and (ii) outlining concrete data and modeling requirements for future work on real-time, event-driven traffic management in smart cities
Predicting Athlete Performance Using Machine Learning Models
No full textThe purpose of this research is to develop and assess multi-modal machine learning for robust performance analysis to predict athletic performance and evaluate injury risk. The study employed Data Analytics approach, where composite features, i.e., Training Stress and Recovery Score,were modelled to characterize training-recovery correlations. It had four different regression models (MLR, RFR, SVR, DNN) and four different classification models (Logistic Regression, RFC, SVM, DNN). DNN exhibited an increased degree of effectiveness in the forecasting Synthetic Performance Score (R2 =0.9998). Notably, the Random Forest Classifier (RFC) turned out to be the most valid predictor of injuries risk (F1-Score 0.7736; AUC-ROC 0.5572) in case the problem of imbalance between classes was dealt with by upsampling. The findings confirm that non-linear ensemble and engineered features can be used to translate physiological and training data into useful and proactive information to coordinate and optimize athletes. Keywords: Sports Analytics, Machine Learning, Random Forest, Injury Prediction, Athlet
Biodegradation of Common Post-Consumer Plastic
No full textPlastic waste is widespread throughout the Laurentian Great Lakes watershed. During environmental exposure, plastic undergoes abiotic and biotic degradation. The presence of novel plastic substrates coupled with rapid microbial turnover may lead to selection for or evolution of metabolic pathways and enzymes with enhanced biodegradation capabilities. This study investigated the biodegradation potential of bacteria isolated from debris accumulation hotspots - stormwater retention ponds, storm drains, and tributaries - in the Lake Ontario watershed within Rochester, New York. Bacterial isolates were exposed to commonly littered plastics for 240 d: cellulose acetate from smoked cigarette filters, high-density polyethylene from takeout shopping bags, polypropylene from chip bags, and expanded polystyrene from packaging foam. Biodegradation was assessed by Fourier-Transform Infrared spectroscopy (FTIR), Scanning Electron Microscopy (SEM), and weight loss measurements. SEM images had biofilm formation on all tested polymers, but surface change results were insignificant. Polymer weight loss and FTIR evidence were polymer- and species-specific, with most dramatic changes on cellulose acetate with Rahnella sp., Serratia sp., Pseudomonas putida, Citrobacter sp., and Escherichia coli. Unfortunately, environmental persistence is present among all of these polymers as degradation of their backbones was not present. Understanding the microbial species and potential rate of degradation for different types of polymers is crucial for assessing the environmental fate of household plastic debris and for future development of plastic-degradation as a mechanism for remediation
A UX Approach to Improving Patient Experience in Healthcare
No full textDoc Ease is a UX research and design project that addresses the emotional and functional difficulties patients experience before visiting a doctor. Through user interviews, competitive analysis, and design iteration, this project identified key user pain points such as unclear appointment processes, lack of trust, and pre-visit anxiety. The resulting mobile app offers clear booking tools, visual doctor information, medication reminders, and an emotional wellness module to help patients prepare with confidence. The project contributes to the field of healthcare UX by integrating functional usability with emotional care, proposing a design framework that centers patient experience in both logic and feeling
Comparative Analysis of Machine Learning Models for Spam Email Detection
No full textThe current research examines one of the most effective approaches to spam email detection based on machine learning and natural language processing (NLP). The study is placed in the context of the rising cyber threats and the influx of emails, where the spam/ham data is to be classified correctfully using the combination of the Logistic Regression, NLP (including tokenization, lemmatization, and TF-IDF vectorization). The questions of the research were devoted to the efficiency of such an approach and the interpretation of its results. The data used are obtained by a publicly available Kaggle data set that contains 5,572 labeled email messages. The quantitative approach was applied, i.e. model training, evaluation, and visualization. The outcomes proved to be very accurate, precise, and had high rates of ROC-AUC proving the efficiency of the model. As conclusions show, when properly preprocessed, Logistic Regression can provide a low-cost but strongly performing method of carrying out spam-detection. Among suggestions, it is possible to calculate ensemble methods, class imbalance, and discover deep learning models in real-time implementation in further studies
Use and benefits of a therapeutic hypnosis interactive media program for chronic pain management: A pilot study
No full textChronic pain is a common condition that has significant negative effects on individuals and society. Opioids continue to be the primary treatment provided to the billions of individuals with chronic pain, despite their general lack of efficacy and significant negative side effects. A growing body of research supports the potential of therapeutic hypnosis for helping individuals better manage their chronic pain. However, access to this treatment is limited. The availability of a hypnosis digital therapeutic has the potential to address this access problem. The current study was designed to evaluate the feasibility of a digital therapeutic hypnosis prototype (not yet released to the broader public). Fifty-four individuals with chronic pain were randomly assigned to have (1) four weeks of access to the prototype or (2) two weeks of no access followed by two weeks of access. Feasibility and clinical outcomes were assessed by the prototype and at baseline and again two and four weeks after randomization. Medium or larger effect size pre- to post-session improvements in the study co-primary outcomes (current pain and feeling inspired; Cohen\u27s d = 0.67 and 0.60, respectively) were observed when listening to a 10- or 20-minute therapeutic hypnosis session. In addition, medium or larger effect size benefits across several important pain-related variables were observed after four weeks of access and use. These preliminary findings support the continued development of the application in order to provide an easily accessible and highly scalable option for people wishing to gain control over their chronic pain
Home
No full textHome is a film where I wanted to explore the idea of the phrase so where s home? being a confusing & spiraling phrase. This is something that hits really close to me. It is a phrase that always seems to catch me off guard. As an international student, I have been around a couple of states for studies. I have made homes in all these places. I have had core memories there that link me to those locations. However, the feeling of not truly belonging would always linger. I really wanted to create a film that captures the main character feeling that inner turmoil of not being able to give a simple answer when asked. Over the course of creating this I have come to the realization that I am a complex person and my answer cannot be simplified and doing so is just a disservice to my background. -Artist Statement, May 202
Predicting Violent Crime Hotspots
No full textThis thesis presents a systematic literature review and an empirical demonstration focused on predicting violent-crime hotspots. Drawing on 50 studies published between 2010 and 2025, the review synthesises methodological developments across hotspot mapping, spatio-temporal modelling, risk terrain analysis, and machine-learning approaches. The review highlights a clear evolution from retrospective density maps to more dynamic, data-driven techniques, while also identifying persistent challenges related to data bias, temporal granularity, environmental context, fairness, and operational implementation. To complement the review, the thesis applies kernel density estimation (KDE) and three ensemble machine-learning models—Random Forest, Gradient Boosting, and XGBoost—to 769,680 geocoded violent-crime incidents recorded in Chicago between 2015 and 2024. Using a strictly temporal hold-out design, KDE achieved strong baseline performance with a Predictive Accuracy Index (PAI) of 2.64 at 5% coverage, while the machine-learning models achieved area-under-curve (AUC) values exceeding 0.93, with Random Forest yielding the highest score (AUC = 0.9365). These results confirm findings from the literature that machine-learning techniques offer improved discrimination over traditional density-based methods, particularly when forecasting short-term micro-scale variation in violent crime. The study concludes that although predictive models have advanced considerably, their practical deployment remains constrained by issues of data quality, spatial-temporal reso- lution, interpretability, and fairness. Effective implementation therefore requires rigorous governance, transparent modelling pipelines, and safeguards to prevent reinforcing historical disparities. Future research should integrate real-time data streams, environmental context, and community-informed evaluation frameworks to ensure more accurate, equitable, and accountable hotspot prediction
Whisper Wall: Transforming Global Superstitions into a Unified, Sound-Reactive Immersive Experience
No full textSuperstitions, often dismissed as irrational in today’s data-driven world, remain deeply rooted in human culture and behavior, offering meaning and a sense of control amid uncertainty. Whisper Wall is an immersive, sound-reactive installation that invites audiences to explore these enduring beliefs through whispered audio narration and symbolic, shadow-inspired visual storytelling. This project re-engages audiences with rich, cross-cultural superstitions, encouraging reflection beyond skepticism to thoughtfully examine their origins and significance. Blending visual communication, motion design, sound, and interaction, Whisper Wall creates a multisensory experience where users activate intimate stories via interactive elements—whisper-like audio delivered through red paper cups coupled with minimalist, evocative posters and animations in negative space. This interface reveals the emotional, cultural, and psychological dimensions of superstitions, highlighting how fear, luck, hope, and control weave through human behavior and tradition. This paper details the conceptualization, design process, and execution of Whisper Wall, demonstrating how storytelling and a carefully crafted visual language can demystify superstitions and foster cultural curiosity. By bridging myth and memory, belief and reflection, the project positions visual communication as a powerful medium to connect the known with the unknown, transforming abstract ideas into tangible, shared human experiences