Descoberta de Conhecimento a partir de Dados das Eleições Municipais das Regiões do Brasil

Abstract

The current Brazilian political situation provides all citizens the right to vote, sustained by a democratic regime characterized by political freedom. This freedom allows multiple parties and candidates, making the elector need to acquire information about competitors, often in the media, or in the environment where they live. After the end of each election are generated a lot of data, which, when summarized, indicate which candidates were elected. But these data are not enough to say why a candidate was elected, i.e., which factors influenced the choice of his name on the ballot. To know which are these factors it is necessary to analyze other types of data beyond those generated on election day. This information is provided by TSE (Supreme Electoral Court), where it is possible to know age, occupation, education level, marital status, and other information about each candidate. Using these data it is possible to find out know how these characteristics influences the final result. For this purpose data mining is a useful tool because it deals with large databases, searching for patterns not previously known. The purpose of this work is to use the J48 data classification algorithm, available at the WEKA software tool to classify data on mayoral elections of the year 2012 in five regions of Brazil by selecting a state of each region. The results show that factors such as political experience of the candidate, campaign spending and age directly influenced the outcome of elections, as well as other factors such as total assets and the level of education of the candidate.