PREDICTIVE ACCURACY AND TRANSFERABILITY OF A NEWLY PROPOSED TRAFFIC FLOW MODEL AND SELECTED MODELS

Over the past eight decades, functional forms of an empirical macroscopic fundamental diagram have been continually proposed yet the predictive power of most models at mid and high densities continuous to be questionable to the traffic flow research community. K-fold cross-validation was used to assess and compare the predictive performance of some macroscopic equilibrium fundamental diagram with a newly proposed model. The results reveal that the newly proposed model perform better than the other models in predicting traffic states, particularly, in mid and high densities. This shows that the proposed model could be useful for extrapolation of traffic states at mid and high densities. Furthermore, the proposed model shows consistently good performance across various roads despite its few parameters. This proves that the proposed model has high flexibility that enables it to adapt to most of the trends of traffic data. It is therefore concluded that a single-regime model with few parameters is capable of accurately describing empirical fundamental diagram.