Prototipo de App móvil que detecte la población en riesgo de reprobación antes del primer bimestre de la UTPL en la modalidad a distancia

Abstract

Abstract: This thesis proposes the design of an innovative mobile prototype, based in a predictive model based on historical data of students in modalities to distance. The primary objective is to anticipate whether a student will pass or fail their subjects during the first two months. The solution seeks to provide alerts to students, offering a valuable tool to identify possible academic difficulties. The CRISP-DM methodology guides the process, using algorithms such as Random Forest, Support Vector Machine and Logistic Regression. The thesis raises specific objectives, such as the creation of a mobile App that alerts students at risk of failure and the application and evaluation of the model predictive. The methodological proposal has a structured approach, while the Deployment of the mobile application model seeks to promote academic success. The Deployment phase culminates with the transformation of the model into a compatible API with the Flutter environment, providing an efficient and accessible solution for early detect the academic performance of students, thus reducing the dropout rate and improving general performance in the university environment.