HLGST: Hybrid local-global spatio-temporal model for travel time estimation using Siamese graph convolutional with triplet networks

Alfateh M. Tag Elsir, Alkilane Khaled, Yanming Shen

Abstract: Highlights•A dynamic composition unit extracts traffic information using geohash and sparse-DTW.•A new hybrid model is proposed to capture local–global spatial and temporal dynamics.•The TCN-SA module captures local traffic variations and relations between features.•The triplet-Siamese GCN module captures global and semantic correlations among nodes.•An adaptive learning algorithm shows our model’s superiority on two traffic datasets.