US Airline On-Time & Flight Delay Statistics by Route (BTS)
Pricing
from $4.40 / 1,000 route climatology results
Historical US flight delay, cancellation and on-time statistics from official BTS data. Rollups by route, airline or airport and month with full delay-cause breakdowns — carrier, weather, NAS, security, late aircraft — honest denominators, cancellation causes and data-vintage on every row.
First month of BTS on-time data to include, as YYYY-MM (e.g. "2026-05"). Leave blank to default to the month before the latest published month. The BTS series begins 1987-10 and is published with a ~2-3 month lag; the run discovers the latest available month live and reports it on every row. An unusable value (slash dates, bare years, month 13) FAILS the run loudly — it is never silently ignored.
Last month of BTS on-time data to include, as YYYY-MM. Leave blank to default to the latest published month. A window spanning more than 12 months is rejected (each month is a ~30 MB BTS download). A month BTS has not yet published is reported as missing — with allowPartial false (the default) the run fails loudly rather than quietly narrowing your window.
Optional. BTS reporting-carrier codes to include, e.g. ["WN","AA","DL","UA","B6"]. Leave empty for all reporting carriers (13 in 2026-05: AA, AS, B6, DL, F9, G4, MQ, NK, OH, OO, UA, WN, YX). A code matching no flights fails the run with the list of carriers actually present — never a silent empty dataset.
Optional. IATA origin airport codes, e.g. ["DEN","MDW"]. Leave empty for all origins.
Optional. IATA destination airport codes, e.g. ["LGA","DCA"]. Leave empty for all destinations.
How to roll flights up. "route" = one row per (origin-dest x month) across airlines; "airline" = one row per (carrier x month); "origin_airport" / "dest_airport" = one row per (airport x month); "airline_route" = one row per (carrier x origin-dest x month). Omitted = "route".
Suppress groups with fewer than this many scheduled flights in a month (noise control — a 2-flight 'route' produces meaningless rates). Suppressed-group count is disclosed on every emitted row as groups_suppressed_below_min_flights. Set 0 to keep everything. Omitted = 30.
Hard cap on emitted rows — each row is one billed result. Enforced by a single audited code path for every groupBy mode; when the cap bites, rows are dropped smallest-groups-first and every emitted row carries results_truncated=true. Omitted = 500.
If a requested month cannot be retrieved (not yet published by BTS, or the download failed), the run FAILS loudly by default so a partial window is never mistaken for a complete one. Set true to proceed with the months that did arrive — every row then reports months_missing and data_complete=false.
