Projects

These are personal and academic projects from shortly after my graduation in 2021. They are kept here as an archive. My current work is described on the front page.

QuickRoll

QuickRoll is a lightweight and efficient dice rolling app aimed at Dungeons and Dragons players. A simple aliasing system allows both players and dungeon masters to save particular rolls, like multi-part attacks with numerous rolls, for quick reuse. The app includes a GUI which bridges the gap between purely text based dice rollers and physical dice.

Designed entirely by myself and written primarily in JavaScript and React, it runs as a static single-page GitHub Pages site.

Live demo · Source on GitHub

Kaggle: readability ease prediction

Correlational heatmap of features extracted from text excerpts — A correlational heatmap of features extracted from the excerpts, such as average words per sentence.

Using NLP techniques learned at university, I submitted to a Kaggle competition where readability ease had to be predicted from given excerpts. Written in Python with pandas, seaborn, and scikit-learn.

Full notebook

Kaggle: heart attack analysis and prediction

Boxen plots of continuous features split by whether the patient had a heart attack — Part of the exploratory analysis: boxen plots for the continuous features, split by the target variable.

Using a heart attack data set published on Kaggle, I applied basic machine learning techniques to predict heart attacks from various medical information. Written in Python with pandas, seaborn, and scikit-learn.

The plots reveal numerous outliers, which matters when scaling the continuous variables. They also push back on intuition: while prior knowledge might suggest that older individuals are at greater risk, heart attacks skew younger in this data.

Full notebook