The US Bureau of Transport statistics provides historic flight schedule and actual flight performance data in CSV tables:

http://www.transtats.bts.gov/DL_SelectFields.asp?Table_ID=23...

But it's cumbersome to work with.

I am working (on and off) on a DBMS benchmark based on this data. As part of that endeavor, I have a script which:

* Automates downloading the CSVs.

* Creates an appropriate SQL database schema.

* Performs a bit of rudimentary cleaning (e.g. invalid character codes; optional)

* Loads the CSV files into the database.

So that, from the command-line, you could get the flight on-time performance data by merely typing in something like:

  /path/to/usdt-ontime-tools/scripts/setup-usdt-ontime-db -r  -db-name ontime --first-year 2019 --last-year 2020

it's available within this repository:

https://github.com/eyalroz/usdt-ontime-tools

the caveat is that, for now, the only DBMS supported directly is MonetDB: https://www.monetdb.org/ , a FOSS analytics-oriented columnar DBMS.

An adaptation of the script for other systems (MySQL/Maria, PostgreSQL) should be straightforward, since the commands are SQL'ish after all. If you're interested in that, open an issue or write me.