View in EDS

Patrones de desempeño docente en Educación a Distancia utilizando aprendizaje automático.

Saved in:
Bibliographic Details
Title: Patrones de desempeño docente en Educación a Distancia utilizando aprendizaje automático.
Alternate Title: Teacher Performance Patterns in Distance Education Using Machine Learning.
Authors: Zarza, Kárilyn Brunett1,2 kbrunettz@uaemex.mx, Valdovinos Rosas, Rosa María2 rvaldovinosr@uaemex.mx, Cármen Montaño Reyes, Patricia del2 pamr@uaemex.mx, Gómez Quiróz, José Jacobo1 jgq2405irc@gmail.com
Source: TE & ET: Revista Iberoamericana de Tecnología en Educación y Educación en Tecnología. Jan-Jun2026, Issue 43, p41-54. 14p.
Subject Terms: *Machine learning, *Distance education, *Data analysis, *Effective teaching, Random forest algorithms, Data mining, Clustering algorithms
Abstract (English): The rapid expansion of distance education has driven the adoption of learning platforms that generate vast volumes of data; however, these data often go unanalyzed, risking the loss of valuable insights. This study profiles virtual instructors on the UAEMéx SEDUCA platform using machine learning within the CRISP-DM framework. A quantitative analysis of 1,041 records and 24 variables extracted from the platform’s teaching performance report was conducted. Preprocessing involved data cleaning, K-NN imputation, and outlier detection. For descriptive analysis, K-means and Hierarchical Clustering was applied and validated with silhouette scores; the predictive phase employed a Random Forest classifier optimized with GridSearchCV and validated via confusion matrix and cross-validation. Three instructor profiles were identified—minimal interaction (15 %), moderate interaction (66 %), and intensive interaction (19 %). The most influential predictors were total graded activities and the percentage of on-time submitted evidence, whereas features such as email exchanges and student count proved less relevant. The model achieved 95.2 % accuracy after optimization. This approach offers universities a reproducible framework for leveraging platform data to inform virtual instructor training and selection. [ABSTRACT FROM AUTHOR]
Abstract (Spanish): El crecimiento acelerado de la Educación a Distancia ha impulsado la adopción de plataformas que generan grandes volúmenes de datos, pero estos no se analizan suficientemente y la información podría perderse. Este estudio perfila al docente virtual de la plataforma SEDUCA de la UAEMéx empleando aprendizaje automático y el ciclo CRISP-DM. Se realizó un análisis cuantitativo con 1 041 registros y 24 variables obtenidas del reporte de desempeño docente de la plataforma. En el preprocesamiento se realizó limpieza, imputación con KNN y detección de outliers. Para el análisis descriptivo se aplicó K-means y Agrupamiento Jerárquico; se validó con coeficiente de silueta; la fase predictiva utilizó Random Forest optimizado con GridSearch y validado mediante matriz de confusión y validación cruzada. Se identificaron tres perfiles: docentes de interacción mínima (15 %), moderada (66 %) y alta (19 %). Las variables más influyentes fueron total de actividades calificadas y porcentaje de evidencias entregadas a tiempo; otras, como correos y número de alumnos, mostraron menor relevancia. El modelo alcanzó una exactitud del 95,2 % tras optimización. Este enfoque ofrece un marco para que las universidades aprovechen datos de sus plataformas en la planificación de la formación y selección de docentes virtuales. [ABSTRACT FROM AUTHOR]
Copyright of TE & ET: Revista Iberoamericana de Tecnología en Educación y Educación en Tecnología is the property of School of Computer Science, La Pata University and its content may not be copied or emailed to multiple sites without the copyright holder's express written permission. Additionally, content may not be used with any artificial intelligence tools or machine learning technologies. However, users may print, download, or email articles for individual use. This abstract may be abridged. No warranty is given about the accuracy of the copy. Users should refer to the original published version of the material for the full abstract. (Copyright applies to all Abstracts.)
Database: Education Research Complete
Description
Abstract:The rapid expansion of distance education has driven the adoption of learning platforms that generate vast volumes of data; however, these data often go unanalyzed, risking the loss of valuable insights. This study profiles virtual instructors on the UAEMéx SEDUCA platform using machine learning within the CRISP-DM framework. A quantitative analysis of 1,041 records and 24 variables extracted from the platform’s teaching performance report was conducted. Preprocessing involved data cleaning, K-NN imputation, and outlier detection. For descriptive analysis, K-means and Hierarchical Clustering was applied and validated with silhouette scores; the predictive phase employed a Random Forest classifier optimized with GridSearchCV and validated via confusion matrix and cross-validation. Three instructor profiles were identified—minimal interaction (15 %), moderate interaction (66 %), and intensive interaction (19 %). The most influential predictors were total graded activities and the percentage of on-time submitted evidence, whereas features such as email exchanges and student count proved less relevant. The model achieved 95.2 % accuracy after optimization. This approach offers universities a reproducible framework for leveraging platform data to inform virtual instructor training and selection. [ABSTRACT FROM AUTHOR]
ISSN:18510086
DOI:10.24215/18509959.43.e4.