A Model Based on Trend-Seasonal Decomposition and GCN for Traffic Flow Prediction

Traffic flow prediction plays an important role in Intelligent Transportation Systems(ITS). To improve the accuracy of traffic flow prediction, this paper proposes a multi-location based on Trend-Seasonal Decomposition and GCN Traffic Flow Forecasting Models for the task of multi-location traffic flow prediction. In this paper, the proposed model mainly consists of two functions: First, the Trend-Seasonal component decomposes the temporal data of traffic flow into a more predictable trend part and a seasonal or periodic part. Second, GCN is used to obtain spatial information between different observation points and improve the accuracy of multi-position prediction. Finally, the experiments for the PeMS04 and PeMS08 data sets are carried out to verify the effectiveness of proposed model.