A New Way to Predict Flight Delays

Coming home for thanksgiving break, I found myself stuck at the Richmond airport for 10 hours more than I had planned for. Our flight had been delayed, and I arrived in Boston at 2 in the morning instead of the original 4 PM. However, recently, researchers from Binghamton University in New York have created a revolutionary computer model that can accurately predict flight delays quickly. This model could potentially save travelers, like myself, hours of time before arriving at the airport.

The current method for predicting flight delays uses artificial neural network computer models, otherwise known as ANNs. Sina Khanmohammadi, leading the study, notes that their "proposed method is better suited to analyze datasets with categorical variables (qualitative variables such as weather or security risks instead of numerical ones) related to flight delays" (ScienceDaily). ANN models contain nodes that take in variables and predict outcomes through analyzing these variables. Through previous data, ANNs can look for patterns and are self- learning. Nevertheless, these networks can also be backfilled with data from previous flights and this can slow the process of flight delay predictions, causing delays to be reported later than desired.

This new research has created a new ANN to help airlines easily determine the relationships between input like weather or previous delays and output- the flight delays themselves. The method analyzes fourteen variables. When the team used the model to predict delays for hypothetical flights expected to land at JFK airport in New York City on January 21st, they found the model predicted delay length with "20 percent more accuracy" and took "40 percent less time" to reach those conclusions (EconomicTimes). They found that previous delays causing a chain reaction of delays were the biggest causes of late arrivals. In fact, that was one of the causes for my own delay last Tuesday. Not only will this new information be useful to fliers, but it can also help airline companies retain their customers and efficiently compensate for delayed flights.

References:
Content:
https://www.sciencedaily.com/releases/2016/11/161114103905.htm
http://economictimes.indiatimes.com/news/science/researchers-find-faster-way-to-predict-flight-delays/articleshow/55432529.cms
https://www.bupipedream.com/news/75742/auto-draft-171/

Images:
http://www.newsx.com/sites/default/files/styles/home_image/public/field/image/Researchers-find-faster-way-to-predict-flight-delays.jpg?itok=uqpIdhIG
http://image.slidesharecdn.com/final-141219125424-conversion-gate01/95/airline-flights-delay-prediction-2014-spring-data-mining-project-3-638.jpg?cb=1418993788