Enhancing Traffic Prediction with Spatio-Temporal Deep Learning: A GCN-LSTM Hybrid Model

This paper presents a hybrid deep learning model that combines Graph Convolutional Networks (GCN) and Long Short-Term Memory (LSTM) networks for spatio-temporal traffic speed prediction. By leveraging the strengths of GCN in capturing spatial dependencies and LSTM in modeling temporal patterns, the model is capable of accurately predicting traffic speeds across a network of sensors. We evaluate the model on two distinct datasets: METR-LA and PEMS-BAY, demonstrating its effectiveness in both short-term and long-term forecasting. Additionally, the introduction of an attention mechanism further enhances the model’s predictive accuracy by focusing on the most relevant sensors.