Análisis de datos sociodemográficos para la toma de decisiones en gestión educativa de nivel superior: caso de estudio graduados años 2018 – 2023

Abstract

This study analyzed sociodemographic data to support decision-making in higher education management in Ecuador, with an emphasis on graduates from 2018–2023. The objective was to identify the factors that influenced university access and to estimate their predictive power for supply planning. An observational approach was used, and the CRISP-DM methodology was followed. Four official sources were combined: high school graduates, university admissions, labor market indicators, and live births, organized by period and province. After cleaning and standardizing the data, K-Nearest Neighbors (k=3, distance-weighted, Euclidean metric) was used to model access. Ten-fold cross-validation was applied, and the data were split into 70% for training and 30% for testing. The results indicated that prior education and group size were the most important factors: the total number of students promoted and births accounted for most of the variance, while labor market indicators had no significance in the model. The performance reached R² = 0.701, MAE = 1,376.50, and RMSE = 2,247.21, with underestimation bias at extreme values and heteroscedasticity at high demand levels. Therefore, the conclusions suggested that planning should focus on retention and promotion in upper secondary education, using demographic-educational projections to determine the number of slots. Furthermore, it is important not to forget that there are unseen institutional and family variables that can affect access peaks. It was suggested that future research include individual data, school quality indicators, and comparisons between provinces, as well as employ mixed-methods approaches that assess cultural impact and academic outcomes, to improve evidence-based decision-making.