TRAFFIC FLOW PREDICTION BASED ON SVM-LSTM MODEL

This paper introduces a novel hybrid model for traffic flow forecasting, leveraging the strengths of both Support Vector Machines (SVM) and Long Short-Term Memory (LSTM) recurrent neural networks. The SVM, LSTM, and the combined SVM-LSTM model were trained using a pre-processed dataset of traffic flow data. The resulting hybrid model enables more accurate prediction of traffic flow dynamics by capturing both temporal dependencies and complex non-linear relationships. Experimental results demonstrate that the SVM-LSTM model significantly outperforms individual SVM and LSTM models in terms of forecasting accuracy. This improved performance highlights the potential of hybrid approaches and the importance of integrating multiple machine learning techniques for enhanced traffic flow forecasting in the context of Intelligent Transportation Systems (ITS).