Machine learning advancements in traffic forecasting: hybrid optimization of LS-SVM for urban traffic management

Economic development has led to a continuous increase in people's demand for transportation, which has led to an imbalance between the supply and demand of road facilities. Traffic congestion, traffic safety and other issues have caused the road traffic environment to become increasingly poor, thereby hindering the development of cities. Therefore, in response to the inability of traditional traffic flow prediction techniques to handle constantly changing traffic flow data, a machine learning traffic flowing predicting model on the foundation of least squares support vector machine is proposed. A hybrid optimizing algorithm based on particle swarm optimization and genetic algorithm is proposed to optimize the parameters of least squares support vector machines in response to the sensitivity of model parameters. Three factors of genetic algorithm are introduced into particle swarm optimization to optimize this model. These experiments confirm that the fitting accuracy between model's predicted values and the actual values is over 90%, and the residual fluctuation is relatively small. Its RMSE is approximately 33.52% to 75.76% lower than the other three algorithms, its MAE is approximately 27% to 67% lower than other algorithms, and its EC is approximately 1% to 5% higher than other algorithms. The proposed hybrid optimization algorithm can find a set of optimal parameters, making the least squares support vector machine have good stability and prediction accuracy in predicting traffic flow problems. This model can be used to predict traffic flow, congestion conditions and other traffic indicators, providing a reliable theoretical basis for urban traffic management.