Evaluation of Real and Virtual Concept Drifts in Flight Delays in São Paulo During Pre, Intra, and Post-Pandemic Periods

• Fabiana Santos Federal Center for Technological Education Celso Suckow da Fonseca (CEFET/RJ)
• Lucas Giusti Federal Center for Technological Education Celso Suckow da Fonseca (CEFET/RJ)
• Diego Carvalho Federal Center for Technological Education Celso Suckow da Fonseca (CEFET/RJ)
• Eduardo Ogasawara Federal Center for Technological Education Celso Suckow da Fonseca (CEFET/RJ)
• Jorge Soares Federal Center for Technological Education Celso Suckow da Fonseca (CEFET/RJ)

DOI: https://doi.org/10.5753/sbbd.2024.243111

Abstract

Flight delays pose significant challenges to operational efficiency and passenger satisfaction. This study investigates the detection of real and virtual concept drifts in flight delays at São Paulo's main airport (SBSP) during the pre-pandemic, intra-pandemic, and post-pandemic periods. Using the Naive Bayes model and integrating data from the Voo Regular Ativo (ANAC) with meteorological information from NOAA, the study assesses the impact of different concept drift detection methods on predictive performance. The analysis reveals that virtual drift techniques, such as KSWIN, were more effective in stable contexts, while basic methods like Passive and Inactive excelled during the pandemic. These results underscore the importance of selecting and combining detection techniques to enhance the accuracy and adaptability of predictive models across various operational scenarios.

Keywords: Flight Delays, Classification, Concept Drift, Virtual Drift, Real Drift

References

ANAC (2023). The Brazilian National Civil Aviation Agency. Technical report, [link].

Gama, J., Medas, P., Castillo, G., and Rodrigues, P. P. (2004). Learning with drift detection. In Brazilian Symposium on Artificial Intelligence (SBIA), pages 286–295. Springer.

Gama, J., Zliobaite, I., Bifet, A., Pechenizkiy, M., and Bouchachia, A. (2014). A survey on concept drift adaptation. ACM Computing Surveys, 46(4).

Giusti, L., Carvalho, L., Gomes, A. T., Coutinho, R., Soares, J., and Ogasawara, E. (2022). Analyzing flight delay prediction under concept drift. Evolving Systems.

Gui, G., Liu, F., Sun, J., Yang, J., Zhou, Z., and Zhao, D. (2020). Flight delay prediction based on aviation big data and machine learning. IEEE Transactions on Vehicular Technology, 69(1):140 – 150.

Iwashita, A. S. and Papa, J. P. (2019a). An Overview on Concept Drift Learning. IEEE Access, 7:1532 – 1547.

Iwashita, R. and Papa, J. P. (2019b). An Overview on Concept Drift Adaptation. Journal of Artificial Intelligence Research.

Lu, J., Liu, A., Dong, F., Gu, F., Gama, J., and Zhang, G. (2019). Learning under Concept Drift: A Review. IEEE Transactions on Knowledge and Data Engineering, 31(12):2346 – 2363.

Moreira, L., Dantas, C., Oliveira, L., Soares, J., and Ogasawara, E. (2018). On Evaluating Data Preprocessing Methods for Machine Learning Models for Flight Delays. In Proceedings of the International Joint Conference on Neural Networks, volume 2018-July.

NOAA (2023). Climate at a Glance Global Time Series. Technical report, [link].

Raab, C., Heusinger, M., and Schleif, F.-M. (2020). Reactive soft prototype computing for concept drift streams. Neurocomputing.

Rani, S. S., Ali, A. I. A., Marie, A., El-Bannany, M., and Khedr, A. M. (2023). Air Traffic Data Analysis Using Recurrent Neural Network (RNN) Classifier During COVID-19. In Proceedings - 17th International Conference on Signal-Image Technology and Internet-Based Systems, SITIS 2023, pages 402 – 408.

Sakthithasan, S. and Pears, R. (2016). Capturing recurring concepts using discrete Fourier transform. Concurrency and Computation: Practice and Experience, 28(15):4013 – 4035.

Teixeira, C., Giusti, L., Soares, J., dos Santos, J., Amorim, G., and Ogasawara, E. (2021). Integrated Dataset of Brazilian Flights. In Anais do Brazilian e-Science Workshop (BreSci), pages 89–96. SBC.

Webb, G. I., Hyde, R., Cao, H., Nguyen, H. L., and Petitjean, F. (2016). Characterizing concept drift. Data Mining and Knowledge Discovery, 30(4):964 – 994.