Hierarchical Reinforcement Learning for Fraud Detection
A hybrid Q-learning and decision-tree system for spotting fraudulent transactions — my CS5841 machine learning final project at Michigan Tech.
Projects
Write-ups of the things I have built and the data I have dug into: machine learning coursework, a two-year scraping pipeline, and a run of data journalism pieces on Scottish education, health and public statistics. Most of them come with the code.
Machine Learning System to Predict Phishing Sites Using HTML, URL & Metadata
Three classical ML models and a CNN, trained on a historical phishing corpus and then tested against phishing sites collected in 2026 — where every model fell apart.
Grade Inflation in Scotland
Or genuine improvement? Trends in top-grade attainment in Scottish Highers, 2009 to 2023.
Strathclyde's Changing Demographics
How gender balance, capacity and student origin shifted at the University of Strathclyde between 2014/15 and 2022/23.
A Two-Year Web Scraping Pipeline
Building, running and monitoring an R and Python scraper that collected public video-ranking data across 31 countries for two years.
How Hot is Scotland? – Fires in the Last Decade
Fire incidents in Scotland have decreased over the past decade. While this is a positive trend, most fires remain ‘Not Accidental,’ particularly in densely populated areas such as Glasgow. This post explores interactive maps, trends, and correlations behind Scotland’s fire statistics.
Smoking in Scotland – How Households Have Quit Cigarettes
An analysis of how smoking rates in Scotland have changed between 2012 and 2019