Undergraduate Research

Permanent URI for this collection

Browse

Recent Submissions

Now showing 1 - 8 of 8
  • Modeling Graduate Student Success Utilizing Admissions Data: A Pilot Study
    (2023) Daniels, Rebecca; McConnell, Brandon M.
    Assessing student success is critical in university graduate school admissions. This is inherently difficult due to the heterogeneous applicant pool, widely varying grading standards across international universities, and other factors. This research is a pilot study to develop mathematical models that leverage admissions data to predict student success in graduate school for a limited set of graduate programs. The goal is a repeatable workflow to support (not replace) graduate admissions committees. Admissions data is matched with academic performance in key courses selected by Directors of Graduate Programs (DGPs). Initial supervised learning models include random forest decision trees and regression to analyze the data and produce useful models, including program-specific scoring models for international school-degree pairs. Initial data exploration led to the construction of multiple models to answer several key questions from DGPs regarding program admissions policies. The findings will highlight factors that contribute to student success and frame potential decisions regarding the admissions process, curriculum changes, and support to improve the likelihood of student success. This proof of concept is a foundation for future research in the study of graduate student success. This research provides a workflow to exploit additional data, DGP feedback, and additional models to support robust decision-making regarding admissions and support for graduate students.
  • NFTs for Trustless Donations
    (2023) Bowman, Noah; McConnell, Brandon M.
    University donations currently require a level of trust that funds are used according to their intended purpose as specified by the donor. Blockchain technologies, to include non-fungible tokens (NFTs), enable the design of new systems featuring transparency, verification, and immutability. Combining this technology with encryption technologies can preserve donor and university confidentiality and compliance where required. In this research, we develop a prototype using the Solidity programming environment to leverage the Ethereum as a decentralized ledger. Our design employs smart contracts to transfer currency and provide donors transparency on the actual use of funds. The design is an initial proof-of-concept to explore how blockchain technologies can enhance current university development processes for charitable giving. Presented at the 2023 NC State Undergraduate Research and Creativity Symposium, Raleigh, NC.
  • Constrained Suitability Optimization of Military Officer Assignments
    (2021) Matthews, Spencer; Fletcher, Matthew; King, Russell E.; McConnell, Brandon M.
    Every year, the US Army spends millions of dollars moving officers between duty stations. There is currently a lack of planning and decision support models available to assist branch managers with managing this talent pool while considering both readiness and cost tradeoffs. We consider the problem of assigning US Army Officers to jobs based on their talent profile. First, we transform and map a dataset of officers and jobs into a usable dataset. Then we introduce a general assignment problem in the form of a mixed integer linear program where talent requirements and budget are constraints. Finally, we optimize the model in order to get the best possible match of officer to job subject to budget considerations. Ongoing research is already underway to support visualizing tradeoffs between overall quality of match and cost, and identifying distinct solution sets for further consideration.
  • Analysis of Organizational Performance at the US Army National Training Center
    (2020) Werner, Jack; Leone, Nicolas; McConnell, Brandon M.; Lawrence, Brandon
    The US Army National Training Center (NTC) could benefit from the use of real-time data analytics to improve the operational effectiveness of its military training operations through an enhanced After-Action Review (AAR). NTC captures Global Positioning System (GPS) data from various sensors during 14-day training rotations that include battle simulations and a live-fire exercise. In this study, we seek to model operational effectiveness for offensive and defensive operations at NTC by identifying and evaluating key measures of performance. Analyzing this data will permit automated models to allow rotating units to more effectively understand their operational effectiveness, identify key areas of improvement, and implement those changes. The data analysis is being performed in R, with the future possibility of creating a real-time interactive dashboard. In order to analyze operational effectiveness, we first create quantitative metrics based on US Army doctrine relevant to NTC operational problem sets. The performance measurements serve as predictors in a regression model to predict operational effectiveness. Fluctuations in friendly Mass and Concentration were found to closely mirror the operational timeline that the unit experienced. Work is currently underway to include the effectiveness metric, which will serve to measure the unit’s overall performance. By conducting real-time data analysis, NTC can improve its ability to provide excellent feedback to rotating units, which will produce much more effective training. The dashboard will provide quantitative results for rotational units to analyze their adherence to doctrinal fundamentals and provide greater context in their performance reviews.
  • Cognitive Performance Analysis of Deployed US Army Unmanned Aerial Surveillance Operators
    (2020) Werner, Jack; Lawrence, Brandon; McConnell, Brandon M.
    The current Unmanned Aerial Surveillance (UAS) Fighter Management Program (FMP) Standard Operating Procedures (SOP) in the US Army may inadvertently contribute to sub-optimal work-rest cycles and ultimately sub-optimal operational effectiveness while deployed. The purpose of this research is to analyze the cognitive performance of deployed US Army UAS operators. The study relies upon observational data consisting of daily sleep data, physiological data, and cognitive test results collected from Soldiers during a 90-day deployment. Descriptive statistical analysis permits visualizing the data over time to identify trends through an operational deployment. Normalized cognitive test data relative to baseline scores allows the development of linear models to isolate significant predictors from the sleep and physiological data sets. While no significant statistical relationships were discovered due to the small sample size, we observe that the majority of Soldiers experienced a decrease in spatial processing, target identification, and memory at or before deployment day 50, as well as some significant decline during the deployment as a whole. These findings can be used to anticipate cognitive decline and support policy recommendations in order to mitigate the risk associated with deteriorating cognitive performance in deployed UAS operators. Description: Poster presented at the 2020 Summer Undergraduate Research & Creativity Symposium, 29 July 2020, NC State University, Raleigh, NC.
  • Readiness Versus Cost Tradeoffs for Additive Manufacturing in a Spare Parts Supply Chain
    (2019) Winz, Ryan D.; McDermott, Kyle C.; Kay, Michael G.; King, Russell E.; Hodgson, Thom J.; McConnell, Brandon M.
    Additive manufacturing (AM) is an emerging technology with disruptive implications in supply chain dynamics that is just starting to be realized. With AM technology, a specific classification of parts can be produced quicker and at lower incremental costs than traditional methods. This technology permits augmenting a spare parts supply chain by allowing downstream locations removed from suppliers to rapidly replace parts that would normally require significant lead times. This research aims to analyze the improvement in readiness and cost that AM could provide for various supply chain designs and for different demand classifications. The research evaluates the impact of AM in a pilot study with three different parts and a supply chain consisting of one traditional manufacturer, three distribution centers, and two service locations. Part demand is generated for each service location and is filled either by AM machines or from the traditional manufacturer via distribution centers. The model explores a total of 32 different network designs which specify the location and configurations of AM at distribution centers and service locations. Part demand is intermittent and classified using average demand interval (a measure of how often nonzero demand periods occur) and squared coefficient of variation (a measure of variance among nonzero demands) to extract best design insights by demand type. A hybrid procedure uses state-of-the-art demand forecasting and a mixed integer linear program to determine an optimal inventory policy for each demand classification. That policy is then fixed in a Monte Carlo simulation that evaluates many possible demand outcomes over a planning horizon to obtain readiness and cost metrics using a frontier-based approach. The two main metrics output by the model are total cost—which encompasses overhead, material, and transportation costs—and total backorders, which measures unfilled demand volume over a single time period. Preliminary results lead us to expect AM will increase costs in the short run, but will increase readiness and decrease long run costs for more erratic part demands. The most important demand characteristic proves to be the volatility of nonzero demand periods, which is measured by the squared coefficient of variation. Scenarios with larger variation have increased backorders and therefore, benefit greatly from the introduction of AM. For parts with predictable demand however, AM leads to very little improvement. The next logical step to employ this model is to utilize it in military logistical situations that have large scale, global implications. In forward deployed scenarios, readiness is essential for all equipment. Additionally, resupplies take even longer and are extremely costly due to the distance from the continental US. The implications of employing AM in these situations would drastically improve the military’s warfighting capabilities. After designing realistic deployed scenarios with multiple stage supply chains, the model would be able to answer where most effectively AM can be constructed and its effects on military logistics as a whole.
  • Determining Evacuation Fleet Sizes for US Noncombatant Evacuation Operations in South Korea
    (2019) Winz, Ryan; Kearby, John; McConnell, Brandon M.
    The purpose of this research is to support US noncombatant evacuation operations (NEO) planning in South Korea and enhance an existing decision-support tool with a simulation model that evaluates alternative resource allocations using outputs from an optimization model. Designed in Simio, the simulation model replicates the South Korean transportation network using nodes and timed arcs. Buses, helicopters, and trains traverse the network with various fleet sizes and corresponding allocation of these vehicles on different pre-determined routes. Evacuees arrive to assembly points following a stochastic time-varying arrival rate for each region. Key outputs of interest include resource utilization, the average number of evacuees at each node, and the total evacuation time. Multiple computational experiments analyzing the scenario of evacuating US Department of Defense families and US government employees reveal increasing the bus fleet size decreases the total evacuation time, with diminishing returns, until a practical limit for the system is reached. Increasing the number of helicopters results in diminishing returns on time saved as well, but without the performance plateau for all realistic helicopter fleet sizes. Military planners could adopt these methods to better assess evacuation operations under different conditions and to determine how resource allocation affects the total evacuation time in a chaotic environment in an effort to adequately support a NEO mission in South Korea. This research enablers planners to better capture tradeoffs between evacuation time and required resources for the commander while providing the capacity to conduct timely what-if analysis for operational risk assessment.
  • Muscle Composition Changes with Brachial Plexus Birth Injury
    (2018) Tamburro, Margaret K.; Fawcett, Emily B.; Dixit, Nikhil N.; Saul, Katherine R.; Cole, Jacqueline H.