Traffic Flow Prediction Using SPGAPSO-CKRVM Model

Received: 5 February 2020 Accepted: 1 April 2020 Traffic flow prediction is popular research of ITS. Traffic flow prediction models based on machine learning have recently been widely applied. In this study, we use machine learning algorithms, heuristic algorithms, and parallelization technology to propose a traffic flow prediction model based on Relevance Vector Machine called Genetic Algorithm and Particle Swarm Optimization based on spark parallelization optimized Combined kernel RVM (SPGAPSO-CKRVM). First, combined kernel functions are constructed based on common kernel functions. Second, a parameter optimization algorithm is proposed to optimize the parameters of combined kernel functions by Genetic Algorithm and Particle Swarm Optimization. To reduce time consumed by the parameter optimization algorithm, we parallel the parameter optimization algorithm by Spark. Finally, the proposed model is verified with the real data of Whitemud Drive in Canada. The experimental results indicate that SPGAPSO-CKRVM has greater accuracy than other prediction models and parallelization technology reduce time consumed by the parameter optimization algorithm significantly.