Abstract:Accurate flight delay prediction is crucial for the secure and effective operation of the air traffic system. Recent advances in modeling inter-airport relationships present a promising approach for investigating flight delay prediction from the multi-airport scenario. However, the previous prediction works only accounted for the simplistic relationships such as traffic flow or geographical distance, overlooking the intricate interactions among airports and thus proving inadequate. In this paper, we leverage causal inference to precisely model inter-airport relationships and propose a self-corrective spatio-temporal graph neural network (named CausalNet) for flight delay prediction. Specifically, Granger causality inference coupled with a self-correction module is designed to construct causality graphs among airports and dynamically modify them based on the current airport's delays. Additionally, the features of the causality graphs are adaptively extracted and utilized to address the heterogeneity of airports. Extensive experiments are conducted on the real data of top-74 busiest airports in China. The results show that CausalNet is superior to baselines. Ablation studies emphasize the power of the proposed self-correction causality graph and the graph feature extraction module. All of these prove the effectiveness of the proposed methodology.

What problem does this paper attempt to address?

The problem that this paper attempts to solve is: **How to accurately predict flight delays in order to improve the safety and effectiveness of the air traffic system**. Specifically, most of the existing flight - delay prediction methods only consider simple relationships between airports (such as traffic flow or geographical distance), while ignoring the complex interactions between airports. Therefore, these methods are insufficient in terms of prediction accuracy. To solve this problem, the author proposes a self - correcting spatio - temporal graph neural network combined with causal inference (CausalNet) to model the relationships between airports more accurately and predict flight delays. The following are the main contributions of this paper and the specific methods for solving the problem: 1. **Introducing causal inference**: Use Granger causal inference to construct a causal graph between airports, thereby more accurately capturing the delay propagation relationships between airports. 2. **Designing a self - correcting module**: In order to reduce the information loss when constructing the causal graph, a self - correcting module is designed, which can dynamically adjust the elements in the causal graph to make it more suitable for the current delay situation of the airport. 3. **Considering airport heterogeneity**: Propose an information extraction method based on graph convolution, which can better handle the heterogeneity between different airports, thereby more accurately extracting spatial correlation information. 4. **Combining spatio - temporal dependencies**: Use long - gated recurrent units (LGRUs) to capture temporal dependencies and further improve the prediction accuracy. Through these methods, CausalNet outperforms the existing baseline models in multiple prediction time ranges, especially when dealing with the delay propagation problem in complex airport networks. ### Summary of Mathematical Formulas - **Calculation of average flight delay time**: \[ y_t^i=\frac{m(t)^i+\rho\cdot c(t)^i}{a(t)^i} \] where \(m(t)^i\) represents the total departure flight delay time of airport \(i\) in the time period \((t - 1,t)\); \(c(t)^i\) represents the number of cancelled flights; \(a(t)^i\) represents the total number of planned departure flights; \(\rho = 180\) minutes represents the equivalent delay time of cancelled flights. - **Definition of prediction task**: \[ \hat{Y}_{t + 1},\hat{Y}_{t + 2},\ldots,\hat{Y}_{t + m}=F(X_{t - r},X_{t-(r - 1)},\ldots,X_t;\theta) \] The goal is to minimize the total loss function \(L(\theta)\), that is: \[ \min_{\theta}L(\theta)=\sum_{t + 1}^{t + m}L(\hat{Y}_t(\theta),Y_t) \] - **Granger causality test formula**: \[ Y_t^a=\alpha_0+\sum_{i = 1}^l\alpha_iY_{t - l}^a+\sum_{j = 1}^l\beta_jY_{t - l}^b+u_{1t} \] \[ Y_t^a=\alpha_0+\sum_{i = 1}^l\alpha_iY_{t - l}^a+u_{1t} \] The existence of causality is tested by calculating the F - statistic. - **Formula of self - correcting module**: \[ HC_t=\alpha\cdot[X_t\|H_{t - 1}]+\beta\cdot\Theta\star G[X_t\|H_{t - 1},C_t] \] \[ \rho_1=\tanh(FC(HC_t)\odot E_1) \] \[ \rho_2=\tanh(FC(HC_t)\odot

A Spatio-Temporal Approach with Self-Corrective Causal Inference for Flight Delay Prediction

Modeling Delay Propagation in Airport Networks via Causal Biased Random Walk

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

Transport causality knowledge-guided GCN for propagated delay prediction in airport delay propagation networks

FAST-CA: Fusion-based Adaptive Spatial-Temporal Learning with Coupled Attention for airport network delay propagation prediction

A spatial–temporal model for network-wide flight delay prediction based on federated learning

Predicting Flight Delay with Spatio-Temporal Trajectory Convolutional Network and Airport Situational Awareness Map

Improving Airport Arrival Flow Prediction Considering Heterogeneous and Dynamic Network Dependencies

A credible traffic prediction method based on self-supervised causal discovery

A Short-Term Traffic Flow Prediction Method for Airport Group Route Waypoints Based on the Spatiotemporal Features of Traffic Flow

Spatio-Temporal Data Mining for Aviation Delay Prediction

Airport Delay Prediction Based on Spatiotemporal Analysis and Bi-LSTM Sequence Learning

Deep Learning Architecture for Flight Flow Spatiotemporal Prediction in Airport Network

Spatio-temporal causal graph attention network for traffic flow prediction in intelligent transportation systems

Wireless Network-Aided Delay Information System Correlation with Airport Grid Distribution Based on Multideterminants Big Data

A Multi‐step Airport Delay Prediction Model Based on Spatial‐temporal Correlation and Auxiliary Features

Spatio-Temporal Contextual Conditions Causality and Spread Delay-Aware Modeling for Traffic Flow Prediction

When Dynamic Causality Comes to Graph-Temporal Neural Network

GT-CausIn: a novel causal-based insight for traffic prediction

Flight Delay Prediction using Airport Situational Awareness Map