The design of a Bayesian Network vehicle traffic flow prediction model for Johannesburg

摘要

Vehicle traffic congestion in Johannesburg has a negative impact on the economy of South Africa in that services and products are not being rendered on time. In this paper prediction models were constructed using historical data and machine learning algorithms to help commuters mitigate traffic congestion. The results show that the Bayesian model provides a reliable alternative for traffic flow prediction as it outperformed the Naive Bayes, K-NN and the Decision tree models. Cross-validation and the rmse were used to evaluate the models. These results will benefit commuters and employers and save costs for companies, improve the South African economy as well as assist Johannesburg in aligning future road traffic strategies.