The paper introduces RSTGCN, a Railway-centric Spatio-Temporal Graph Convolutional Network that predicts average arrival delays for all incoming trains at a specific station during a given time period. It incorporates train frequency-aware spatial attention and other architectural innovations to improve predictive accuracy. Experiments on a newly released dataset covering 4,735 Indian Railway stations show RSTGCN outperforms state‑of‑the‑art baselines by 18% in MAE, 14% in MAPE, and 1–8% in RMSE.
By Koyena Chowdhury, Paramita Koley, Abhijnan Chakraborty, Saptarshi Ghosh
The paper presents a machine‑learning approach to predict train delays in Finland using the Finland Integrated Train‑Weather (FI‑TW) dataset, which merges railway operational records with data from about 200 nationwide weather sensors. Three feature sets were tested with XGBoost at Oulu central station: full weather features, instant observations only, and derived weather categories. The category‑based features—hierarchical classes such as Blizzard, Heavy Snow, and Extreme Cold—yielded the best performance, achieving an R² of 0.78, RMSE of 8.5 min, and MAE of 3.7 min, an 11 % R² improvement and 10 % error reduction over the other configurations.
By Vinicius Pozzobon Borin, Jean Michel de Souza Sant'Ana, Nurul Huda Mahmood
arXiv:2601. 16592v2 Announce Type: replace-cross Abstract: Train delays result from complex interactions between operational, technical, and environmental factors.
By Vinicius Pozzobon Borin, Jean Michel de Souza Sant'Ana, Usama Raheel, Nurul Huda Mahmood
arXiv:2608.20980v1 Announce Type: new
Abstract: Graph neural networks (GNNs) are routinely employed for short-range forecasting on multivariate time series with a spatial graph structure. Despite the...
By Kenneth Martin, Simon Heilig, Asja Fischer, Michel F. C. Haddad, Adam M. Sykulski, Moshe Eliasof
The paper presents an empirical benchmark of nine modern deep‑learning models for time‑series forecasting of smart‑meter energy consumption, evaluated on two publicly available datasets. It examines how historical input length, prediction horizon, and model architecture affect accuracy, finding that longer historical context improves performance up to a saturation point and that accuracy declines with longer horizons. The study also compares computational complexity, showing that lightweight architectures achieve similar performance to heavier models, and notes that model choice has limited impact across most demographic and household subgroups.
By Behnaz Kavoosighafi, Maria Eidenskog, Wiktoria Glad, Katerina Vrotsou
arXiv:2608. 13023v1 Announce Type: new Abstract: Relational Deep Learning (RDL) models multi-tabular databases as temporal heterogeneous graphs to enable end-to-end representation learning.
By Jakub Pele\v{s}ka, Gustav \v{S}\'ir