This project started as a way to quantify heartbreak — turning a failed relationship into a dataset. It began with a pile of old text messages, neatly exported as CSV files — the emotional equivalent of a spreadsheet full of red flags. I converted these to JSON and spent far too long learning regex, metacharacters, NLP, and Rita.js, slowly wrangling over 300,000 lines of raw data into something I could actually make sense of.

What I found was both predictable and unsettling. It turns out the slow death of a relationship is statistically visible if you know where to look — word count drops, less mentions, and the steady decline of frequency.

In the end, I turned this digital debris into a series of 6 printed data visualizations — a kind of relationship autopsy for the chronically introspective. If you’d like to see the whole, mildly traumatic process, check out my Medium post in the link above.