Digital Transactions Mining to Characterize Temporal Rhythms of a City

摘要

The identification of crime hotspots by its spatial and temporal components has become a widely used strategy by police intelligence for crime reduction [1] [2]. However, this kind of approach is possible thanks to a large amount of granular data these entities have at their disposal, something not always present in developing countries such as Ecuador. With that in mind, the following study seeks to implement transaction records from the city of Guayaquil in association with criminal records to define temporal and spatial patterns for crime hotspots. We first describe our popularity metric as a function of these transactions for each circuit of the city. Then, we find the correlation between this metric and several crime indicators such as theft and murder. Finally, we establish relations between commercial density and criminal records for each circuit. Our results show that these transactions records can indeed be used to establish possible crime hotspots, albeit just for December 2016. Additional information about police outposts - and open data variable - could lead to more accurate results.