Clustering Socioeconomic and Spatial Patterns of School Dropout

Abstract

Academic dropout is a fundamental problem at all educational levels. Several studies have been conducted to identify the characteristics that motivate students to drop out. Few studies have found patterns related to location. The objective of this article is to show the geographic patterns of academic dropout of undergraduate university students and to analyze them to improve decision-making in Peruvian universities. The methodology for this purpose is KDD (Knowledge Data Discovery), commonly known as data mining, which includes the Clustering technique and the application of the K-means algorithm, one of the most popular and efficient algorithms. The results reveal notable geographical patterns, including the university site and the district to which dropout students belong. The results help to identify the university campuses with the highest dropout rates and the districts from which the dropouts come. The latter is analyzed from a socioeconomic perspective.