Analysis of homicides and basic food basket cost time series from Ilhéus and Itabuna cities, Brasil

Abstract

Violence is among the biggest problems faced by Brazilian public administration and the increase in the number of homicides in interior cities is an alarming factor for the performance of public security institutions. Understanding its causes is a major challenge for social development and analyzing the available data on homicide indicators can be an efficient method that results in appropriate conclusions. This work aims to apply computational techniques to analyze historical data series on homicides and also on the cost of the basic food basket in the cities of Ilhéus and Itabuna, Bahia, Brazil, for later application in prediction models with deep learning. Data were collected from the public databases of the Mortality Information System (SIM) and the Basic Basket Cost Monitoring Project, at the State University of Santa Cruz. Data were processed and transformed into univariate time series, which underwent exploratory analysis, in order to extract statistics and characteristics such as trends and seasonality and also test them for stationarity. It was also verified the existence of correlation between the series. The results showed the non-stationarity of the series, with verification of stochastic and deterministic trends, and in addition, the correlation values between the series of homicides and the cost of the basic food basket proved to be irrelevant.