Application of Spatial Data Science on Results of the Saber 5 Test

Abstract

The purpose of the Saber tests is to improve the quality of education in Colombia through periodic evaluations of the competencies development in basic and secondary education students. Although different data mining studies of the Saber tests’ results have been carried out, not many are based on spatial analysis of data in the educational field. This article presents a study based on exploratory and spatial analyses of the 2016 Saber 5 test dataset, which aims at identifying the relationship between the results of the test and the geographic or spatial distribution of the data. For the development of this research, four methodological phases were established: data selection and collection, correlation analysis, spatial data analysis, obtaining conclusions. Using the free GeoDa tool, a quantile and regression analysis was performed on the results of the Language, Mathematics, and Natural Sciences areas. Similarly, an analysis based on clustering was performed using the KMeans algorithm. The results obtained made it possible to determine the linear and spatial correlation between the data and the different departments of Colombia at the level of the three areas considered, which may be of interest in decision-making processes by the Ministry of Education. In addition, the research allowed to verify the usefulness of the GeoDa tool for conducting studies based on exploratory and spatial analysis in different application contexts.