Airline delays are an ongoing challenge for the airline industry, affecting both airlines and passengers. By detecting and understanding outliers in the airline data, particularly related to delays, the industry can improve operational efficiency, optimize schedules, and enhance customer satisfaction. Outlier detection plays a key role in identifying unusual patterns or disruptions in flight delays that could indicate underlying issues at airports, with aircraft, or within the airline’s operations.
Delays can result from a variety of factors such as weather conditions, air traffic, or airport-specific inefficiencies. Identifying unusual delay patterns early allows airlines and airports to take preventive actions. By using outlier detection techniques, the goal is to find outlier airports with significant deviations in average arrival delays, whether due to excessively long delays or unusually early arrivals.
This workflow accesses a sample of data from the airline dataset and detects outlier airports based on the average arrival delay in them. The techniques applied are numeric outlier, z-score, DBSCAN and isolation forest. The outlier airports detected by each of these techniques are visualized on a map of US using the Geospatial Analytics Extension for KNIME.