Ever been frustrated while traveling (yes, of course you have). Knew that Delta was probably a better bet than Spirit, but succumbed to the very low drip pricing of Spirit? Wondered whether it was worth it to wake up at 4AM for an early flight? Or, perhaps worst of all, had a flight delayed for two hours or more (NOT due to weather), or even canceled?

I think most people, especially on this website, have a nightmare travel story.

The shrewd of us check FlightAware or flightstats to check the historical delay data of a flight. While this gets us 60% of the way there, it's not perfect because it works with Flight Number as the "key" (to use hash table parlance). But flight numbers are shuffled around often—flight XYZ42069 might leave at 7PM on a Friday (terrible, TERRIBLE time to fly) but used to leave at 7AM on a Tuesday (great time to fly), lulling the shrewd traveler into a false sense of security.

FlyOnTime cuts through all of that by only looking at the five variables that seem to matter: airline, origin airport, departure hour, day of week and month of year.

FlyOnTime does not use any "fancy" methods such as "machine learning" because it is not needed in this use case — five dimensions in, one dimension out. We are not dealing with images, videos, text or audio. It is a simple extension that could have been built weeks after Chrome Extensions were launched in 2009.

The plan is to keep it free for a long time. If people actually use it, I can then think about making a premium version for power users (travel agents and executive assistants). This would perform complex calculations for connecting flights, as well as incorporate an element of the time value of money (for example, a CEO could probably justify spending 3x more to reduce the probability of massive delay by even 7%, because her time is worth so much). Definitely open to ideas and collaboration though. Like I said, my sole goal for the time being is to empower travelers to know what they're buying better than before.

Methodology is here: https://flyontime.substack.com/p/flyontimes-methodology

Disclaimer: Only works for US flights. I have scoured the web for EU data, which is usually pretty good, but have been unable to find it how I need it (at the flight level with fields for flight date, origin, destination, departure delay). Please please please let me know in the comments if you know of this dataset's existence as I'd love to extend it to Europe.

user:	BiologyBum
created:	June 16, 2019
karma:	3