Shannon - Entropy - Based Artificial Intelligence Applied to Identify Social Anomalies in Large Latin American Cities

Abstract

The emergence of social anomalies in developing countries have demanded to use alternative methodologies that allows us to identify concrete problems that to some extent constitute a negative factor that substantially delays both social and economical progress of a country either in the middle or long term. Because most of the social factors that would stop such progress falls entirely in the territory of the social dynamics particularly in that large cities, concretely in this paper we apply the Log to the Shannon's entropy as a kind of tool to identify in parallel the level of risk for street criminality as well as the presence of traffic chaos in a large city. For this end we use an acceptance-rejection-based algorithm that selects one geographical square of a certain zone belonging to Lima city in Peru. While all squares have same probability to be selected we introduce a memory-based factor that accounts previous criminality-traffic events in some specific areas. Our results have indicated that those dual points criminality-traffic are strongly correlated with social, urbanity, and economic development factors. Simulations from stochastic algorithms have yielded a matching between model and official data of a 85±5%. Therefore the results of this paper are along the direction of the recovery of the main social-economic parameters of Latin American countries by which are the main cause of the apparition of these social anomalies.