Desigualdades educacionais no brasil: uma análise por clusterização de indicadores educacionais e desempenho escolar

Abstract

Educational inequalities in Brazil represent a persistent challenge, reflecting socio-economic and regional disparities that impact school performance. This study proposes an analysis of these inequalities through the clustering of educational indicators and school performance, using unsupervised machine learning techniques, with a focus on the K-Means algorithm. The study seeks to identify patterns and trends in the quality of Brazilian public education over the years 2015, 2019, and 2021. The methodology involves data collection and cleaning from the Instituto Nacional de Estudos e Pesquisas Educacionais Anísio Teixeira (INEP), exploratory analysis, transformation and normalization of indicators, as well as the application of clustering techniques to group municipalities with similar educational profiles. The validation of the clusters was carried out through metrics such as the silhouette coefficient and the elbow method. The results reveal a positive evolution in the Índice de Desenvolvimento da Educação Básica (IDEB), but also highlight regional disparities and persistent challenges, such as the age-grade distortion rate. The analysis of the evolution of the clusters over time showed significant migrations between groups, suggesting changes in educational and socio-economic conditions. The study concludes that clustering is a valuable tool for the formulation of educational public policies, contributing to equity and the improvement of education quality in Brazil.