Application of spatial data mining based on aggregation techniques to detect tra-ffic congestion levels in the city of Trujillo

Abstract

The analysis of vehicular traffic in the road network of the city of Trujillo detected critical points with the greatest influx of vehicles, determining a computational model (Figure 8) using random values to detect vehicular traffic congestion in the road network of the city of Trujillo. For the management and analysis of spatial data mining, the CRISP-DM methodology was used, applying unsupervised learning, because the data are random and simulated in real time, which allowed the use of classification clustering techniques with k-means and dbscan algorithms, and as a result the spatial location of the nodes with a level of congestion was obtained, deducing the possible incidences or eventualities that occur in them at vehicular level, as is the case of speed, time, capacity at each node. The k-means and dbscan clustering algorithms were compared to validate the computational model, using the k-means and Silhouette averaging techniques, respectively.