Two years.
Four continents.
One dataset.
TravelNet is a personal data collection and machine learning platform built to run unattended across 2–3 years of international travel. GPS traces, health metrics, finances — all in one place.
Departure countdown
Philadelphia, PA → Global
What's in here
Data. Analysis. Insights.
Every module of TravelNet feeds into a unified dataset — built to be explored, visualised, and learned from.
Journey
Interactive GPS traces from six countries, rendered in Kepler.gl. Flight arcs, movement density, and time-animated paths across the full trip.
Coming September 2026Data Explorer
Live database statistics, interactive charts, and spending breakdowns by country, currency, and category — updated as I travel.
Coming 2026ML Insights
Travel leg segmentation, fitness trend decomposition, spending pattern detection — machine learning models trained on two years of personal data.
Coming late 2026Open Dataset
An anonymised, privacy-fuzzed subset of the full dataset. GPS coordinates offset by ~1 km Gaussian noise. Free to explore and build on.
GitHub →The route
Six legs. 3 years.
Built with
The stack
A Raspberry Pi 4B runs unattended, collecting data 24/7. Everything streams into a single SQLite database.
Raspberry Pi 4B — always-on server
FastAPI — data ingest endpoints
SQLite — unified travel database
Overland + iOS Shortcuts
Apple Watch Ultra + Health Auto Export
Revolut · Wise · exchangerate.host
scikit-learn · HDBSCAN · pandas
Kepler.gl · Plotly · Dash
Tailscale · Cloudflare R2