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 visually appealing GUI which bridges the gap between purely text based dice rollers and physical dice. A demo of the app is available here and the open source GitHub repository is here.

The app is designed entirely by myself, and written primarily in JavaScript and React, and runs as a static single-page GitHub Pages site.

Using some NLP techniques learned at university, I submitted to a Kaggle competition where readability ease had to be predicted from given excerpts. See the full notebook here. The notebook uses python, pandas, seaborn, and scikit-learn.

The figure to the left is a correlational heatmap of various features that were extracted from the excerpts, such as average words per sentence.

Using a heart attack data set published on Kaggle , I performed basic machine learning techniques to predict heart attacks given various medical information. See the full notebook here. The work is written in python, primarily using pandas, seaborn, and scikit-learn.

The sample figure is part of EDA, showing boxen plots for the continuous features, split by the target variable: 1 if the patient had a heart attack and 0 otherwise. In particular, the plot reveals numerous outliers in the data, which is useful to consider when scaling the continuous variables.

Additionally, while intuition and prior knowledge might suggest that older individuals are at greater risk for heart attack, the plot indicates that heart attacks are skewed younger in this data.