Flight delays could disturb passengers' travel plans and cause financial loss to the airlines. Therefore, efforts have been spent in the past to conduct flight delay prediction, which could assist passengers, airports and airlines in better planning. Existing methods focus on predicting whether a flight will delay and how long the delay will be, but lack an effective explanation revealing possible reasons causing the delay, which is useful for timely decisions. Motivated by the demand of such cause-aware flight delay prediction, we propose an explainable multi-task framework to predict not only the delay duration but also the delay causes. The proposed framework consists of three modules: (1) flight record encoder that derives record embeddings, (2) flight sequence encoder integrating useful signals from other related flights, and (3) flight delay predictor that outputs delay duration as well as the possible causes of the delay. The proposed framework is evaluated on three real-world datasets. The results show that the proposed model outperforms multiple baselines on the flight arrival delay prediction task and is able to provide the causes behind the flight delay.


    Access

    Check access

    Check availability in my library

    Order at Subito €


    Export, share and cite



    Title :

    Explainable Multi-task Flight Arrival Delay Prediction


    Contributors:
    Wang, Tianqi (author) / Lin, Lei (author) / Gao, Jing (author)


    Publication date :

    2021-09-19


    Size :

    972042 byte




    Type of media :

    Conference paper


    Type of material :

    Electronic Resource


    Language :

    English



    Flight Arrival Delay Prediction Using Deep Learning

    Sharma, Nishant / Vijayalakshmi, S. | IEEE | 2024


    Flight Arrival Delay Prediction Using Supervised Machine Learning Algorithms

    Alla, Hajar / Moumoun, Lahcen / Balouki, Youssef | Springer Verlag | 2021




    Research on flight arrival delay prediction based on support vector machine

    Hu, Peng / Hou, Jia / Zhang, Jianping et al. | IEEE | 2022