48440 research outputs found
Sort by
Predicting Stock Prices: A Mathematical Analysis of Large-Cap Pharmaceutical Firms
No full textThe world of investing is characterized by large firms with extensive data and models that are inaccessible to the broader public. Over the past decade, several retail investing companies have emerged; however, the public is still largely unaware of the fundamentals of the companies they invest in and the models that large firms use to select stocks. This paper develops and tests two technical models (a Fourier-based trend model and a stock-index blended model) and evaluates usefulness for retail investing. The models show potential but limited predictive power without refinement. In parallel, this project improves the financial literacy of readers by explaining the methods, code, and limitations
Project 1: Enhancing Earthquake Preparedness in Istanbul
No full textİstanbul, Türkiye is one of the most earthquake prone megacities, situated along the North Anatolian Fault. With a major seismic event expected before 2030, the city faces challenges in urban planning, public awareness, infrastructure resilience, and policy coordination. This project evaluates İstanbul's Earthquake Preparedness through fieldwork, expert interviews, spatial data analysis, and policy review. It integrates historical risk trends, municipal plans and social and psychological dynamics of affected communities, especially post 2023 earthquakes. Our findings show gaps in municipal coordination, limited public trust, and out of date spatial data. We conclude with targeted recommendations for İstanbul Metropolitan Municipality (IBB), and other organizations. Recommendations focus on GIS integration, educational reform, structural monitoring, and inclusive emergency communication strategies
Designing a Multimedia Interactive View of the Thames Tunnel Bazaar
No full textDigital interactives are an enticing way to increase visitor engagement and accessibility for museums. Our project’s goal was to develop a digital interactive for The Brunel Museum's website, displaying the Thames Tunnel bazaar from 1843 to 1865. We met with museum experts to learn best practices for developing interactives, interviewed Brunel Museum staff and volunteers to identify their needs for an interactive, and developed a prototype of the digital interactive. We recommend the museum continue to develop and use this interactive to enhance their engagement with younger audiences, increase accessibility to the museum’s knowledge, and expand the educational resources the museum offers
CS/DS: MQP: AI-Powered Species Identification: Combating Illegal Wildlife Trade with Melting Curve Analysis
No full textIllegal wildlife trade (IWT) continues to pose a major threat to biodiversity, conservation efforts, and global ecological stability. Traditional identification techniques, including morphological analysis and expert interpretation of genetic data, are often impractical in field scenarios, particularly when only partial or processed biological materials are recovered. In this project, we present a machine learning-based pipeline that enhances species identification using melting curve analysis; a genetic technique that detects unique DNA denaturation patterns. Our system integrates a Vision Transformer (ViT) model, optimized through transfer learning and hyperparameter tuning with a python library named Optuna, to classify melting curve data transformed into image representations. Due to the limited availability of labeled samples, especially for rare species, we developed synthetic data augmentation strategies to bolster minority class representation. Our final model achieved 95.35% classification accuracy, significantly outperforming a baseline Convolutional Neural Network (CNN). To make the model accessible to non-specialists and enforcement agents, we built a cross-platform mobile application that interfaces with a remote server to deliver real-time classification results. This work demonstrates the potential of AI-powered molecular forensics in improving wildlife trade enforcement and species conservation by accelerating and automating species identification from genetic data
Artificial intelligence (AI) Detection of Algae in Microscopy Images for Water Source Environmental Management
No full textIn Hong Kong, harmful algal blooms (HABs) have led to severe environmental and economic consequences, worsened water quality, and an increase in fishkills. In our research, our group focused on the causes of HABs, their ecological impacts, and the current efforts to monitor and reduce their occurrence. With our research, we developed a user-friendly website designed to consolidate the data we gathered and educate the public on the topic of algae and water pollution. Our website is also set up to utilize artificial intelligence (AI) to improve algae detection, as it offers promising advancements in efficiently and accurately identifying algae from microscopy images
Specialty Fiber Probes for Light-Matter Interactions and Their Applications in Physical, Chemical, and Biomedical Research
No full textThis dissertation presents the design, development, and application of specialty fiber-based optical systems engineered to probe light–matter interactions across physical, chemical, and biomedical domains. By leveraging the flexibility, miniaturization, and robustness of fiber optics, three distinct platforms are demonstrated to address unmet needs in environmental sensing, laser-tissue interaction, and in situ spectroscopy. First, a novel fiber-based optical trapping platform was developed to manipulate and characterize airborne and liquid-borne microparticles. This system integrates both radiation pressure and photophoretic forces to enable contactless trapping and mechanical probing in open-air and aqueous environments. Through quantitative modeling and experimental validation, the platform provides localized measurements of particle stiffness, viscosity, and drag, enabling real-time mechanical analysis with applications in aerosol science and soft matter physics. Second, a miniaturized fiber-delivered laser surgery probe was engineered for high-precision thermal ablation and tissue characterization. The probe design incorporates topographical imaging and Raman spectroscopy to analyze laser-tissue interactions in biologically relevant scenarios. This work expands the theoretical understanding of laser ablation beyond metal-based models into a biological, multiphysical framework that incorporates heat transfer, tissue deformation, and chemical response, offering a foundation for future clinical integration in robotic and minimally invasive surgery. Third, a multi-fiber Raman sensing system was developed for real-time chemical detection and phase analysis in food and biomedical applications. The compact fiber-bundle design integrates filtering, enhanced signal collection, and spectral calibration strategies to enable sensitive detection of analytes such as acrylamide and to monitor crystallinity transitions during thermal processing. Experimental validation in both static and dynamic environments highlights the system's robustness and clinical potential for optical biopsy and intraoperative diagnostics. These contributions advance the use of fiber optics as multifunctional platforms for in-situ and non-invasive analysis. The outcomes of this work offer new pathways for deploying light-based sensing in real-world environments—ranging from industrial processing lines and biomedical operating rooms to airborne pathogen detection systems—while also deepening the fundamental understanding of complex light–matter interactions in soft and heterogeneous materials
Development of an Ionic Wind Propelled Aircraft - Propulsion
No full textThis project presents the use and application of ionic wind propulsion as a secondary propulsion method in an aircraft and an associated mission. The goal was to identify a mission, and an associated aircraft that would benefit more from ionic wind propulsion than in other circumstances. This could allow the effects of ionic wind propulsion to be more noticeable in the presence of a stronger primary propulsion system. Simulations involving COMSOL Multiphysics and experiments on various electrical and mechanical components of the aircraft were performed to allow an optimal construction of the aircraft considering the practical limitations of ionic wind propulsion. The power systems on board the aircraft must provide enough voltage to generate an electric field strong enough for corona discharge. The mission for the aircraft must occupy a low-speed regime given the small thrust limit. The aerodynamic airfoils of the aircraft were modeled in SolidWorks and were tested for aerodynamic lift, drag, and interactions with corona discharge. The project found simulation results that indicated regions of maximum space charge density and ionic wind velocity were located to the left of the emitter. This is in contrast to the default simulation where both the emitter and collector radii were smaller. Experiments found that decreasing the distance between collectors increases thrust density. Using airfoil shaped collectors resulted in non-conclusive results and using a positive corona discharge results in a better thrust density as opposed to a negative corona discharge. Fai
Mass Timber Bridge Design and Construction
No full textCross laminated timber (CLT) is a type of engineered wood that has been rising in popularity since its invention in the 1990s for its strength and aesthetic qualities. Despite this, current standard plans for timber bridges fail to include it. This project aims to prove the viability of CLT as a cost efficient and sustainable alternative to more traditional bridge design methods like steel and concrete. To do this, structural analysis and comparisons to cost, emissions, and schedule were performed
Design and Analysis of a Constellation of CubeSat Radio Telescopes for the Imaging of Saturn’s Rings
No full textDue to the Earth’s atmosphere, observing extraterrestrial objects and bodies is ineffective when using ground-based telescopes, as the gases in our atmosphere can distort and scatter signals. A satellite in orbit would avoid this issue and be able to capture images without interference. This project presents the design of a constellation of CubeSats to image various astronomical objects using a method similar to Very Long Baseline Interferometry (VLBI). This report describes the design and analysis of the structural and propulsion systems. The structure of the CubeSat was designed using the requirements laid out by the CubeSat Design Specification guidelines, launch vehicle requirements, and subsystem constraints. Vibration analysis was used to ensure the CubeSat stayed within axial and lateral load limits set by the launch vehicle. SpaceX’s Falcon IX was selected as the launch vehicle. A custom folding mesh antenna was designed using SolidWorks – inspired by the antenna used on NASA’s 2018 RainCube mission – with a Cassegrain design to allow the electronic system to be placed behind the dish. The dish design fits within a cylindrical volume, and when deployed it has a diameter of 0.5 meters. ECAPS’s 22 N thruster was selected for the CubeSat to perform the necessary orbit raise, perturbation compensation, and deorbiting maneuvers. Low complexity and high thrust were the two primary factors in the down-select process. This CubeSat design and analysis can be used as a starting point for the development of in-space VLBI telescopes which could improve overall image quality in radio telescopes
The Design and Fabrication of Novel Guitar Frets and Fretboards for Manufacturability and Wear Resistance
No full textManufacturing of traditional guitar fretboards requires extensive craftsmanship. While extensive knowledge exists on fret installation and repair, frets still wear down and require replacement. We designed and prototyped a unified fret-spine design that improves assembly and fretboard construction, while ensuring long-term durability and repair. The fret-spine slots into the fretboard through dovetail joints, allowing for easier removal. We investigated manufacturing methods for small and large-scale production. Investment casting proved feasible for small scale manufacture, while extrusion-based manufacturing could work on a larger scale. We devised a fretboard design that utilized additive manufacturing to accommodate the fret-spine geometry, and compared wear, aesthetic and mechanical characteristics of potential fretboard materials. To simulate wear from guitar playing, we designed and constructed a wear tester. We identified potential fretboard materials for use with the fret-spine design and recommend further exploration of materials and manufacturing methods to support greater accessibility to lutherie and guitar manufacturing