Development of Dynamic Multi-Interval Traffic Volume Prediction Based on System Approach Using Historical and Real-Time Data

The objective of this study is to introduce a dynamic model to estimate multi-interval traffic volume using historical and real-time traffic volume data. This study was brought about by the drawbacks of the existing single-interval prediction techniques, which have been widely applied for the estimation of future state. This paper also includes the applicability of the proposed model using real-world data. The developed model is based on the Nearest Neighbor Non-Parametric Regression (NN-NPR) using real-time and historical data, which are collected by the Toll Collection System (TCS) and managed by the Advanced Data Management System (ADMS) respectively. In an empirical study with real-world data, the presented multi-interval prediction model performed effectively in terms of prediction accuracy to the degree of the application of real ITS systems.