Análise exploratória dos microdados do Censo da Educação Superior: ênfase em cursos de computação

Abstract

Computer science education has gained prominence in recent decades, driven by the growing demand for qualified professionals and the field’s recognition as one of today’s leading scientific disciplines. Despite these advances, there are challenges related to qualifications in computing, such as high dropout rates in these courses. In this context, Educational Data Mining enables the analysis of large volumes of student data to support strategic decisions, provide frameworks for decision-making, and optimize educational processes. The quality of data mining research results is related to the reliability of the database used, which may contain inconsistent or missing entries due to errors in data acquisition. Given this scenario, this study performed the sanitization and exploratory analysis of microdata from the INEP Higher Education Census, focusing on Computing programs in Brazil, which are well-established for use in data mining. Through this exploratory analysis, it was possible to identify inconsistencies in the official database. During this work, these inconsistencies were corrected through the application of data preprocessing and sanitization techniques. As a contribution, this work provides two sanitized and documented versions of the database, aiming to support studies in classification and mining of educational data. Furthermore, it is believed that the applied methodology can serve as a reference for database sanitization in other contexts, contributing to improving the quality of data used in research.