DCU Expo 2024 - Final Year Projects
48 54. Investigating the Expected Goals (xG) Football Metric UsingMachine Learning This project investigates the expected goals (xG) metric used in football data analysis. Using StatsBomb’s in-game data as the foundation, machine learning techniques were employed to decode its calculation, incorporating key features of shot attempts deemed crucial in analysing teamperformance. Insights into teams’ performances are revealed based on the quality of chances they created. Assessing the accuracy of the metric developed in this project was achieved by comparing its xG outputs with both StatsBomb’s xG values and actual match scores. The analysis not only reveals the role of ‘luck’ in football but also scrutinises the concealedmetric of xGwithin football match analysis. Student Programme Data Science Project Area Data Analytics, Statistical Analysis, Deep Learning Project Technology Excel/VB, Matlab, Python, Machine Learning, Tableau, Git Student Name(s) Cara O’Boyle | Ruairi Ryan Email cara.oboyle6@mail.dcu.ie | ruairi.ryan67@mail.dcu.ie Supervisor Prof Alan Smeaton 55. Neuromotor Assessment Device and Kinematic Analysis of ArmMovement This project investigates real-time kinematic armmovements using a neuromotor assessment device (NAD). The NADwas made up of two sensors, located on the wrist and the upper arm, to communicate with each other and find the location andmovement of a kinematic arm. The use of two sensors with a high refresh rate makes this device more accurate and efficient at finding the orientation andmovement of a kinematic arm, which is crucial for the scope of the project. The movement of the kinematic armwas also displayed on a screen for the user to examine. The data output from this device could be used for biomedical purposes to spot early signs of neurodegenerative disorders such as Parkinson’s disease at an early stage, which is paramount to providing effective treatment. Student Programme Mechatronic Engineering (Year 5) Project Area 3-DModelling, Arduino, Augmented Reality, Biomedical Engineering, Data Analytics, Gaming, Graphics, Information Retrieval, Mechanical Design and Manufacture, Mechatronic Systems, Motion Analysis, Power Electronics, Robotics, Sensor Data, Sensor Technology, Simulation Project Technology C/C++, Java Student Name(s) Gokmen Ozturkmen Email gokmen.ozturkmen2@mail.dcu.ie Supervisor Dr Conor McArdle 56. Evaluting the Effects of Automatically Added Early Exits on the Performance of Convolutional Neural Networks Early Exits have been amethod of improving the runtime speed of neural networks for years, but they have needed to be manually added to a network, requiring work to decide where an exit should be placed and how that exit show be structured. This project aims to analyse whether these exits can be automatically added to any convolutional neural network without affecting the overall accuracy of the network while lowering the time taken to run the model. Student Programme Electronic and Computer Engineering (Year 5) Project Area Artificial Intelligence, Computer Vision, Software Development Project Technology Python, Machine Learning Student Name(s) Stephen Condon Email stephen.condon5@mail.dcu.ie Supervisor Dr Robert Sadleir
Made with FlippingBook
RkJQdWJsaXNoZXIy MTQzNDk=