Pragmatic data-driven AI-based approach for exploring the effects of important constructs on flight cancellations in disruptions and infectious diseases

This study presents a pragmatic real-world approach for exploring flight cancellations in disrupted situations and infectious diseases. The airline industry has been prone to disruptions, and a huge number of flight cancellations have been reported in recent years. In this study, an AI-based, data-driven approach is used to explore flight cancellation patterns in the wake of disruption. The correct flight information is important to save time, money and resources for airline customers. Data pertaining to the event of flight cancellation in disruption has not been properly presented in previous studies. Thus, in this study, the prediction power of artificial intelligence tools has been exploited for flight cancellations. Artificial intelligence tools, namely two-class neural network regression, two-class decision forest, a two-class locally deep support vector machine, and two-class logistic regression, were utilized for prediction. The precision, accuracy and recall performances of all the methods used are 100%. The confusion matrix depicts that all four methods correctly predict the outcome of flight cancellation. For the total of 549169 tested flights, 549168 flights have been correctly predicted. The only false positive instance shows that there is only one case where the flight was not canceled, but the methods incorrectly predicted it as canceled. Thus, for almost all the cases, the predictions are correct. The correlation calculated from the huge data of 2745847 flights reveals that flight cancellations have a frail positive relation with departure and arrival delays. Moreover, there is a fragile negative relationship with airtime, weather delay, carrier delay, security delay, NAS delay, and late aircraft delay. The lessons learned from planning and coping with disruptions can be used as preemptive preparation guidelines for handling emergency situations in the future.