Spatial-Temporal Traffic Speed Prediction on Thailand Roads

Abstract

The ultimate goal of our research mission is to build a traffic model for travel time prediction in Thailand in order to improve the mobility domain of the smart city. To achieve our mission, this piece of research places important on a traffic speed prediction of any reference points at an incoming time that is one significant part of the travel time prediction. In this study, we employ a linear model for predicting traffic speed of some kilometer stones in the next several minutes. Our prediction model performs less root-mean-squared-error score under some spatial-temporal conditions. In addition, the temporal-lagged associations among kilometer stones, which were extracted during the feature selection process, are observed as a traffic-dependent network on roads for analyzing the traffic congestion spreads in the future.