Extração de conhecimento nos dados da Universidade Federal do Ceará via mineração de dados: descoberta e análise dos perfis dos alunos

Abstract

Some changes have occurred in the evaluation system for admission to higher education in Brazil. Perhaps one of the most striking changes has been the implementation of the National Examination of Secondary Education (ENEM), as the only way of admission in several public universities, including the Federal University of Ceará(UFC). In this context, it is important to know the profile of the students who enter the university every year. This knowledge can be obtained using the university's own data. As well as large companies and organizations, the university also generates data that without the analysis process have limited meaning. The data can be used as feedstock in the application of knowledge-discovery. Knowledgediscovery in extracting useful knowledge from a large amount of data. Among its various stages, which will be mentioned in this term paper, Data Mining is one of the most important and consists of a set of techniques to analyze large amounts of data trying to find a consistent pattern, summarize the data, extract knowledge or make predictions. It is possible use some of these techniques in the university data. In this study, the technique chosen is clustering, which is based on grouping similar items into groups, and elements of different groups are less similar, each group forming a profile of student of the Federal University of Ceará, seeking to respond to questions about measures taken by the university over the years , such as adoption to SISU and the new frequency rule adopted in the second half of 2008.