Author(s)

VINÍCIUS DE OLIVEIRA LIMA, JOSÉ MARCOS V. A. GOIS

Abstract

Motor vehicle collisions (MVC) are a major problem in Brazil, the yearly number of deaths caused by MVC in the country is comparable to those seen in war zones around the globe. One of the first steps to deal with this problem is using appropriate tools to study it. The Civil Police of the Federal District (PCDF) is the official authority in charge of forensic analysis of MVCs that resulted in deaths or injuries. Each MVC attended by a forensic expert from PCDF generates a considerable amount of information including address, GPS coordinates, street description, type of accident, damages, vehicle identification, and others. All this data is stored in a JSON database. This work describes an interactive platform created by a forensic expert and an undergraduate student using Microsoft Power BI to extract data (through SQL queries) and generate reports from this database. These reports provide meaningful graphs such as geographical heat maps, bar plots regarding MVCs, the road speed limit and the distribution of accidents per month. When observing the first 6 months of 2020, we are able to notice three main spots of MVCs that required a reconstruction analysis. We studied each case within the three spots and a pattern for each one appeared. With this information, we provided solutions for traffic engineering in order to prevent further similar accidents. Therefore, with such insights, we proved our system can give support to government agencies to adapt their policies and introduce new ones to help curb the number of deaths and injuries caused every year due to MVCs in Brazil.

Keywords

forensic, traffic, crash collision, accident, reconstruction, data mining