Classification of Days in the National Airspace System Using Cluster Analysis

Abstract

Scientific methods can describe the National Airspace System (NAS) in ways that provide intuitive insights into its operation and performance. One such method is classification and analysis of historical data. In this study we identify key metrics representing the NAS as a whole, and use cluster analysis techniques to classify days in the NAS spanning a four-year time period. Data are analyzed and compared before and after the September 11, 2001 national tragedy. Through classification, we reduce this data into manageable and meaningful subsets. Each subset has dominant characteristics that exemplify typical behaviors in the NAS, primarily based on traffic volume and weather. The data are then analyzed within and between subsets in order to gain information and knowledge from an otherwise unwieldy superset. The results of such an analysis can be utilized for efforts such as the testing and validation of NAS simulations, NAS trend analysis, cost/benefit annualization, and quality assurance.