Link Prediction and Maximum Flow in Transportation Network

This study conducted link prediction analysis and maximum flow analysis, which provide critical insights into alternative route inferences and traffic flow, based on real transportation network data. The dataset used in the analysis was specifically generated for this purpose. Data collection involved Bluetooth vehicle counting devices installed at 54 intersection points in the city center of Malatya, Turkey. The methodology leveraged approximately 50 million vehicle transition records to weight the transportation network graph. The Ford-Fulkerson method was utilized for the maximum flow analysis, while the Jaccard similarity metric was employed for the link prediction analysis. The graph construction and all analysis processes were carried out using the R programming language and the igraph graph library. The results of the analyses provided significant insights into alternative route corridors within the transportation network and the maximum traffic capacity of the roads. Consequently, the findings enabled the identification of critical points and potential congestion areas. The outcomes are expected to make a substantial contribution to enhancing the efficiency of the transportation network and improving traffic management strategies.