Prediction of Flight Delays Using Bayesian Networks and Data Mining

Understanding and mitigating flight delays is a major long-term objective of the Federal Aviation Administration (FAA). The purpose of this research is to develop a dynamic flight delay model using Bayesian networks and data mining methods. This model represents the dynamics and complexity of NAS through causal paths of flight delay incidents in which conditional interdependences between causal variables are presented in a directed acyclic graph of nodes and arcs, and a conditional probability table. The model is developed through machine learning algorithms, and then evaluated and tested using FAA databases.