A Deep Learning Approach for Flight Delay Prediction Through Time-Evolving Graphs

Flight delay prediction has recently gained growing popularity due to the significant role it plays in efficient airline and airport operation. Most of previous works consider single-airport scenario, which overlooks time-varying spatial interactions hidden networks. In this paper, flight problem is investigated from a network perspective (i.e., multi-airport scenario). To model time-evolving periodic graph-structured information network, approach based on graph convolutional neural (GCN) developed paper. More specifically, regarding that GCN cannot take both time-series structures as inputs, temporal block Markov property employed mine patterns delays through sequence snapshots. Moreover, considering unknown occasional air routes under emergency may result incomplete inputs for GCN, an adaptive embedded into proposed method expose Through extensive experiments, been shown outperforms benchmark methods with satisfying accuracy improvement at cost acceptable execution time. The obtained results reveal deep learning have great potentials problem.