Sistema recomendador de material educativo mediante el análisis de seguimiento en Moodle.

Abstract

The main challenge in virtual education on the Moodle platform lies in selecting appropriate educational material, given the diversity of students’ knowledge and learning styles. To address this complexity and the lack of direct student feedback, a recommender system was developed that analyzes Moodle logs to determine the most suitable material. This project was carried out using the SCRUMBAN methodology, which consists of seven well-defined phases, from goal setting to final product deployment, ensuring accurate and thorough implementation. It was programmed with attention to detailed specifications during development, and extensive testing was performed to ensure functionality, performance, and security requirements were met. The precision measurement results showed a positive consistency, exceeding 89%, while the sensitivity registered a minimum of 68%, still considered acceptable. The overall accuracy reached 76%, indicating satisfactory overall system effectiveness. Furthermore, the low Root Mean Square Error (RMSE) value of 0.96 suggests the efficacy of the system's recommendations, which contributes to its usefulness and reliability. The development of the recommender system was based on data analysis tools such as NumPy, Pandas, and Scikit-learn in Python, as well as Flask, to adapt the recommendations to individual learning needs by incorporating a multiple intelligences test. Although the results reflect a high accuracy, between 94% and 98%, opportunities were identified to improve the identification of significant elements and refine the accuracy of the suggestions, highlighting the system's potential for future improvements.