Sistema inteligente de predição de evasão acadêmica: desenvolvimento e integração de modelo de aprendizado de máquina no curso de licenciatura em computação

Abstract

This paper presents the development of an intelligent system designed to predict academic dropout in the Licentiate Degree in Computing at the Federal Institute of Piauí, Teresina Zona Sul Campus. Higher education dropout is a complex problem driven by socioeconomic, institutional, and pedagogical factors. The general objective was to build a web platform integrated with an artificial intelligence model capable of identifying risk patterns and presenting alerts to educational managers. A qualitative approach supported by software engineering was adopted, with the system organized into modules: a Django back-end, responsible for processing and storing the data, and a JavaScript front-end, with Axios and Chart.js, responsible for the interface. The development of the predictive engine evolved from an initial heuristic version, based on manually defined weights, to the implementation of an artificial intelligence model. The experiments assessed data storage, dashboard rendering, and the contingency mechanism, culminating in a controlled that confirmed the superiority of the artificial intelligence by correcting the extreme mathematical results generated by the weight-based model. It is concluded that, although the retraining of the model with real institutional data is in its final stage, the predictive and software architecture was successfully validated through a synthetic database, resulting in a robust, transparent prototype structurally ready to act in the prevention of academic dropout.