Identificação de regiões populares a partir de trajetórias

Abstract

This paper describes a process of identification of popular regions, through taxi drivers trajectories, where the points where the car is stopped are identified. The data used here belong to the TaxiSimples application and represent 7 days of official taxi drivers, from June 2016 in the city of Fortaleza-CE. Here, two approaches are evaluated to reach this objective, the first strategy is the SMoT, proposed by Alvares et al. (2007a) and is based on intersection between previously mapped trajectories and points of interest, these points were mapped by Google and extracted by mean of a script that uses an API made available by it. The second one is the CB-SMoT, proposed by Palma et al. (2008) and uses a variance of the DBSCAN algorithm to figure out where the car is at low speed to identify regions of interest. DBSCAN was also used for the clustering of the stops, with the objective of identifying clusters of popular regions, measuring them by the number of stops existing near the colors of the clusters. At the end of this, the Script was used again to show points of interest near the popular regions.