Web Platform for the Predictive Analysis of Student Dropout in a Higher Education Institution in Latin America (#1998)

Authors

Blanco Lopez, Jaime

Galeano Ospino, Saray

Niño Manrique, Jhon Fredy

Ramírez Chiquito, Alexander

Gonzalez Santamaría, Andrés Esteban

Abstract

Student dropout in Colombian higher education is a complex problem that affects educational quality and the socioeconomic development of the country. This research focused on the development and implementation of a web platform to analyze and predict student dropout in a denominational higher education institution in Colombia, using a logistic regression model. The research was developed in four phases: 1) literature review and expert consultation to identify input variables related to dropout, 2) requirements definition, system architecture design and technology selection, 3) project development using Kanban methodology, and 4) implementation of the solution, execution of a pilot test and satisfaction evaluation through a survey. The predictive algorithm, with an accuracy of over 90%, allows the institution's professionals to detect early on students at risk of dropping out. The web platform, developed using Python provides an intuitive interface to visualize the current dropout status of faculties and programs, predict student dropout risk and understand the influencing factors. Validation through a pilot test and a satisfaction survey confirmed the effectiveness and usability of the platform, although areas of improvement for future implementations were also identified.