Anomaly detection of private jet flights

i built a small system that tracks a fixed cohort of business jets (from FAA registry data, matched via ICAO hex) and counts how many are airborne at a given time.

it ingests ADS-B exchange heatmap files (30 min intervals), backfills historical counts, and compares the current value to a baseline for similar times of day/week.

the output is a simple deviation, shown as an "emergency level".

in practice it mostly shows a strong daily cycle, with occasional spikes during holidays, or on april 6th when trump threatened "a whole civilization will die tonight".