A series of standalone open-source data analyses. Each takes a public dataset with at least a century of depth, runs serious ML analysis, and presents the results through world-class visualization.

Open Data · ML/AI Analysis · MIT Licensed · 10 Entries
Live 01
UFO Witness Reports
112,000 NUFORC witness narratives treated as a sociolinguistic corpus. Vocabulary tracks culture. Independent witnesses converge. 30 archetypes emerge.
111,961 reports 118 years 10/10 validation
Live 02
American Names
Every name is a cultural wave. 104,000 names across 144 years. Rise, peak, decay — the shape encodes cultural information invisible to people living through it.
104,819 names 144 years 12/18 triggers
Coming 03
Major League Baseball
The game changed. The data shows exactly how and when. Dead Ball through Statcast — every era as a distribution shift.
Coming 04
Extreme Weather
Daily station-level climate data. Every place has a worst day. The app finds it and tells you what it felt like.
Coming 05
American Homicide
The 1990s crime drop is one of the most dramatic shifts in any dataset. Almost nobody knows what the 1970s actually looked like.
Coming 06
Disaster Declarations
Some counties have been declared disaster areas 40+ times. The map of repeat exposure is not the map most people carry in their heads.
Coming 07
Immigration
The 1924 cutoff shows up as a hard wall in the data. The 1965 act shows up as a detonation.
Coming 08
American Prisons
The 1980s show up as a vertical line. Some states look like different countries. State trajectory clustering reveals divergent paths.
Coming 09
Aviation Incidents
The fatality rate in 1930 vs. today barely fits on the same chart. The remaining risks cluster in patterns that are immediately visible.
Coming 10
Epidemics & Outbreaks
From the 1918 flu to COVID. How fast institutions respond — and how that response lag changed over a century.