1st Place Winner of Fall 2019
by Richard Ohr, Siddharth Surapaneni, and Faye Park
Abstract: For this project we are creating a machine learning model utilizing flight data from the US Bureau of Transportation Statistics. In order to use our findings to benefit the community around us, we decided to focus on flight delays and cancellations for flights traveling out of Virginia. Our main goal is to analyze how long flights were delayed and how these delays relate to other attributes about the flight. Given any set of flight information (such as airline, date, time of day, airport), our model will output an estimate of how long that flight will be delayed. The project will also be fitting a model to estimate the probability that the flight will be cancelled entirely. We will modify the data, choose models, and tune parameters to make the model as accurate as possible. This project is mainly geared towards being a practical application. It will let travellers know ahead of time, for any given flight, an estimate of: A: How long the flight will be delayed after its scheduled departure time B: The likelihood that the flight will be diverted or cancelled entirely The benefit of this is twofold. First, travellers will be able to choose flights with minimal delay and risk of cancellation. Second, travellers will be able to accommodate their plans to the approximate delay. This is a significant utility to those who travel by plane, either rarely or frequently.