Abstract
To address the issue of traffic congestion, a new unsupervised incremental learning strategy has been proposed to identify and profile traffic congestion in metropolitan areas. The proposed model can effectively analyze and anticipate traffic situations in urban areas and improve traffic efficiency. Additionally, a clustering method based on Density-Based Spatial Clustering of Applications with Noise (DBSCAN) and Ordering Points To Identify the Clustering Structure (OPTICS) has been suggested to cluster accident-prone regions. This method outperforms other algorithms based on synthetic and actual datasets. Furthermore, this research aims to determine and forecast the traffic flow of the road network using GPS data from floating cars. The traffic condition of the metropolitan road network is determined using an emerging hot spot analysis tool to look for diverse patterns of hot spot formation. Using the time series clustering approach, the road network is partitioned into groups with comparable spatiotemporal features. The three-time series forecasting models are also applied to estimate traffic operation status, and the proposed model outperforms the existing methods. Finally, this study proposes efficient and effective methods for managing traffic congestion in urban areas. These methods can help identify, assess, and forecast traffic congestion levels, crucial for improving traffic efficiency and ensuring public safety.
Keywords: Adjusted Information, Affinity Propagation, Clustering Structure, Density-Based, Hotspot Analysis, Spatial Clustering.