Sistema de predicción del rendimiento académico de matemáticas en instituciones educativas públicas utilizando el algoritmo Support Vector Machine
Average rating
Cast your vote
You can rate an item by clicking the amount of stars they wish to award to this item.
When enough users have cast their vote on this item, the average rating will also be shown.
Star rating
Your vote was cast
Thank you for your feedback
Thank you for your feedback
Authors
Sarmiento Zegarra, Maria PazAdvisors
Espejo Villaizan, Daniel DerekIssue Date
2023-11-13Keywords
Rendimiento estudiantilAprendizaje automático
Modelo predictivo
Clasificación binaria
Student performance
Machine Learning
Predicting model
Binary classification
Metadata
Show full item recordOther Titles
Prediction system for mathematics academic performance in public educational institutions using the Support Vector Machine algorithmDOI
http://doi.org/10.19083/tesis/673418Additional Links
https://audio.com/raupc/audio/11585Abstract
No detectar a tiempo a los estudiantes que puedan presentar dificultades en Matemáticas, puede poner en riesgo la aprobación de dicha materia y el aprendizaje correcto de las competencias necesarias para el futuro desarrollo profesional. Este trabajo propone un Sistema para predecir el rendimiento del estudiante antes y durante el curso en 2 clases “aprobado” y “desaprobado”. El sistema está basado en un enfoque de trabajo de 3 fases. La primera fase es el Análisis y selección de datos, donde luego de la revisión a la literatura se seleccionó 3 conjuntos de variables: el historial académico, las notas actuales y los factores extraacadémicos; estos últimos fueron validados por expertos en piscología educacional. En la segunda fase se diseñaron 3 modelos de predicción con la herramienta Azure Machine Learning. Finalmente, en la tercera etapa se integró los modelos con un sistema web que permite explotar la información relacionada a las predicciones. Los resultados al entrenar los modelos demostraron que la “educación del padre”, el “nivel de interés en la materia” y el “apoyo de la familia” fueron los factores extracadémicos más importantes, obteniendo el Accuracy de 88.7%, 85.5% y 77.4% en los modelos 3, 2 y 1 respectivamente. Además, en el estudio experimental se obtuvo una coincidencia del 80% entre los resultados del sistema versus los resultados de un docente a una misma muestra de estudiantes.Failure to timely identify students who may have difficulties in Mathematics can jeopardize their approval of the subject and proper learning of the necessary competencies for future professional development. This work proposes a system for predicting student performance before and during the course in two classes: "approved" and "failed". The system is based on a 3-phase working approach. The first phase is Data Analysis and Selection, where after reviewing the literature, three sets of variables were selected: academic history, current grades, and extra-academic factors; the latter were validated by experts in educational psychology. In the second phase, three prediction models were designed using Azure Machine Learning tool. Finally, in the third stage, the models were integrated with a web system that allows the exploitation of information related to the predictions. When training the models, the results showed that "father's education," "level of interest in the subject," and "family support" were the most important extra-academic factors, obtaining an accuracy of 88.7%, 85.5%, and 77.4% in models 3, 2, and 1, respectively. Additionally, in the experimental study, an 80% coincidence was obtained between the system's results and those of a teacher on the same sample of students.
Type
info:eu-repo/semantics/bachelorThesisRights
Attribution-NonCommercial-ShareAlike 4.0 Internationalinfo:eu-repo/semantics/openAccess
Language
spaEmbedded videos
ae974a485f413a2113503eed53cd6c53
http://doi.org/10.19083/tesis/673418
Scopus Count
Collections
Except where otherwise noted, this item's license is described as Attribution-NonCommercial-ShareAlike 4.0 International


