Conjunto de dados para análise da evasão acadêmica no curso de licenciatura em computação do IFPI-CATZS

Abstract

Academic dropout is a multifactorial phenomenon that undermines the performance of Higher Education Institutions, generating institutional deficits and the waste of public resources. In the Computing Teaching Degree program at the Federal Institute of Piauí – Teresina Zona Sul Campus (IFPI-CATZS), historical figures reveal dropout rates frequently exceeding 60%, highlighting the urgency of investigating its causes. In light of this scenario, this study aimed to develop a structured dataset to support the analysis of academic dropout in the aforementioned program. The research was based on the Knowledge Discovery in Databases (KDD) process, with emphasis on the selection and pre-processing stages. The socioeconomic and academic data of the incoming students were extracted from the SUAP and Q-Acadêmico systems and integrated into a two-dimensional model, using the Python programming language, the Pandas library, and the Google Colab environment. After collection, the data underwent cleaning, treatment of null values, and encoding, normalization, and standardization techniques. As a result, different datasets were built, one containing only real data and another with values imputed by the mode, totaling 504 instances, which made it possible to identify the predominant profile of graduating and dropout students. It was found that imputation by the mode aggravated disparities in some attributes. It is concluded that the datasets produced represent a consistent basis for future applications in data mining and machine learning, providing support for institutional policies to address dropout.