The goal of this project is to train a model that would allow the user to input details about a potential flight (pre-existing or not) and predict if their flight departure is meaningfully delayed (>15 mins) or not. Flight delays cost the US economy around $30 billion annually, and up to 37% of this amount is composed of costs to individual passengers.

While most flight delays are inevitable, given the scale of this industry even small improvements can have significant impacts. Such a tool can to be useful to delay-sensitive customers, both in terms of airline/airport selection, but also for small-change differentiation - perhaps the 10am flight option is a lot better than the 12pm one. On the airline-level this can provide a way to filter for the one which are not and may be able to be improved, or also when studying new potential routes and schedules which ones would likely have high delay rates.