Please use this identifier to cite or link to this item: https://hdl.handle.net/10419/300321 
Year of Publication: 
2024
Series/Report no.: 
Texto para Discussão No. 3015
Publisher: 
Instituto de Pesquisa Econômica Aplicada (IPEA), Brasília
Abstract (Translated): 
The Ministry of Health's Mortality Information System (SIM) is one of the main sources of information for measuring violent deaths and, in particular, homicides in Brazil. However, between 1996 and 2021, there were 294,752 violent deaths in which the cause was not defined (or 8,7% of the country's total violent deaths). The aim of this article was to estimate the proportion of these deaths that were caused by homicides. The methodology adopted in the work was based on a supervised learning model (machine learning), in which the probabilistic pattern of personal and situational characteristics for each type of event - whether homicide or suicide/accidents - was learned. Our estimates indicated that in the period there were 128,567 hidden homicides, or homicides not officially registered as such. According to our findings, an average of 4,492 homicides went unreported each year. This problem seems to be more serious in four Federal Units: São Paulo, Rio de Janeiro, Bahia and Minas Gerais.
Subjects: 
machine learning
homicide underreporting
Mortality Information System (SIM)
Brazil
JEL: 
C18
C81
K42
Persistent Identifier of the first edition: 
Creative Commons License: 
cc-by Logo
Document Type: 
Working Paper

Files in This Item:
File
Size





Items in EconStor are protected by copyright, with all rights reserved, unless otherwise indicated.