Data science aplicado a dados abertos do Governo Federal: estudos de caso sobre a economia dos municípios brasileiros.

Abstract

The process of analyzing open databases in recent years has gained considerable prominence in the Brazilian scenario since the granting of Law 12,527 / 2011, which guarantees access to public information, allowing for better transparency of public spending by society. Allied to this, numerous discussions arose around the use of Brazilian government microdata, among which we highlight the discussions on social security reform and the analysis of fiscal health in Brazilian municipalities through social security approaches. Thus, this work focuses on the use of Data Science, specifically in the KDD process to analyze microdata from Brazilian municipalities. Thus, in this work, two different approaches are made, the first of which performs a descriptive statistical analysis without inferences, to understand the fiscal health of Brazilian municipalities between 2010 and 2017, through transfers from the RGPS. The second approach to fiscal analysis using the STVAR model through the following variables: expenditure, revenue, and GDP of the municipality of São Paulo. The results of analysis I show that municipalities with populations greater than 100 inhabitants do not show a deficit due to the difference between municipal collections and transfers from the RGPS. In analysis II, the results found show that the economic cycle analyzed when undergoing exogenous shock (or external impulse) can generate changes in the states of recession and expansion with an average duration of 12 months.