Projects
Some of my software projects undertaken during my time as a Masters Student at UCD.
NYSeeNow
I was the Coordination Lead of a team of 6 for my summer research project. The app we developed, NYSeeNow, is a travel itinerary planner and guide, with a focus on utilizing the live busyness of Manhattan to guide users on where to visit. I was largely involved in the Data Analytics and Machine Learning of this project. I also contributed heavily to implementing front end features and overall project management. The frontend relies on React.js and Node.js, with the backend utilizing Spring Boot and Flask. The prediction models were created using Jupyter Notebooks, Scikit-learn, pandas and NumPy. I thoroughly enjoyed this summer project as it opened my eyes to the flow one could find in researching, gathering, cleaning and then actually using real-world data. Learning how to manage a team of 6 provided a great learning experience.
Dublin Bikes
Involved in Full-Stack development of a predictive web app harnessing data from OpenWeather, Google Maps, JCDecaux Bikes. Focused on the model training to predict availability of bikes, but also involved heavily in backend development using Javascript, MySQL, AWS, EC2 and Nginx.
Covid-19 Data Analysis
Analyzed Covid-19 data from the CDC to determine the impact of certain factors on the mortality of people who tested positive for Covid-19. 3 different models were evaluated using different metrics, with Linear Regression having the least predictive power and Random Forest with the most predictive power. A large number of the features in the dataset had almost no impact on the case mortality. The age group of each person had the strongest predictive power by far across all models and feature sets. This makes sense as it is widely known that older people were more susceptible to Covid-19, while younger people would have to be extremely unlucky to die from Covid-19.
The image below shows the feature importance for the Random Forest Classifier Machine Learning Prediction Model:
