Web Application Based on Machine Learning for the Prediction and Prevention of Low Academic Performance in Primary School Students in the District of Matara, Cajamarca

Gallardo C.D.C.; Chulluncuy L.V.T.; Quintana C.A.A. · Proceedings of the IEEE International Conference on Big Data, BigData (2025) · DOI: 10.1109/bigdata66926.2025.11401747

Resumen

This study presents a machine learning-based web application designed to predict and prevent poor academic performance among primary school students in Matara, Cajamarca. Using academic, behavioral, and socio-emotional data, the system identifies risk patterns and supports timely pedagogical interventions. Three models-Naive Bayes, Neural Networks, and Random Forest-were evaluated, with Random Forest achieving the highest accuracy (93.6%) and F1-scores across all classes. The backend was implemented with FastAPI and MySQL, while the frontend was developed in Angular to ensure accessibility for teachers. The tool enables automated monitoring and data-driven decision-making, empowering educators to act proactively. By integrating predictive analytics into rural education, this solution contributes to more equitable, efficient, and informed teaching practices.

Instituciones: Universidad Peruana de Ciencias Aplicadas