Raidell Avello Martínez and Marcos Carrera Ramos
Universidad de Cienfuegos
Learning Management Systems (LMS) are the main computerized educational systems today. In particular, Moodle is one of the most widely used LMS today, which was adopted by the Cuban Ministry of Higher Education to deploy and use in the universities of the system. Like other LMS tools, Moodle collects a large amount of data including all the ratings and interactions of the participants allowing the application of a complete set of Learning analytics (LA) techniques.
The relevance of LA is palpable given its increase in scientific publications and international projects. It is impossible to ignore these techniques as something fundamental embedded in the learning process and in higher education, to manage actions to improve learning. The large volume of data (records) of each participant, from the interaction of hundreds of students, with more than a dozen qualifying activities, show the need to apply specialized tools in LA.
Moodle does not include visual learning analysis functionalities in its modules that it installs by default. There are functional plugin-based extensions to Moodle, which support visual exploration of student activity and performance. But the decision to make a new extension available to teachers depends on multiple factors of the academic institution: size, organization chart, economy, etc. That are beyond the scope of the teacher’s decision (Ji et al, 2020).
There are several pluings (Luna et al, 2017) and external tools (Sáiz-Manzanares et al, 2021) to analyze and monitor the participants of a course in Moodle, including advanced Machine Learning analyzes to predict the performance of students, classify them according to different variables extracted from their interaction and completion of activities among other analyzes (Ahmed & Khan, 2019). However, few of these applications analyse the reports provided by Moodle, that is, they work by connecting directly to the server and extracting the information online.
The goal of the Moodle Offline Analytics web application is to provide a simple and basic solution for analyzing Moodle course logs and providing useful interactive graphs to describe student activity.
Moodle Offline Analytics is a web application for teachers or managers, developed in python and using the Django framework, which analyzes the records of one or more Moodle courses. Faced with more common solutions, such as Moodle plugins or blocks, a web application lacks the bureaucratic, technical and economic problems of installing and updating versions on an institutional server decision (Ji et al, 2020). This application follows an open source and free use policy, under the MIT license, with the latest version available in the GitHub repository (https://github.com/ravellom/moodle_csv_prof).
In order to use the application, it is necessary to enter the course in question and download the records as explained in the help section of the application. Then in the “Load data” section you can upload the file and exclude participants who do not want to be part of the analysis, as well as set the date range to analyse.
Figure 1. Load data and filters
In the general analysis section there is a description of the accesses for the days analysed, a heatmap that provides the accesses by day of the week and time using color intensity to indicate more or less number of accesses, a graph with the amount of access by type of activity and another by specific activities, that is, the most popular activities. Finally, a world map is provided with a color scale according to the amount of access by country.
Figure 2. General analysis
In the participants section, 2 graphs are shown, one with an analysis of the number of participants for the main activities of the course and a cluster analysis that classifies the students into two groups according to the variables presented in the table below of the graphics. This table has 5 variables that describe the contributions of the participants in the main knowledge-building activities that Moodle provides.
Figure 3. Participants analysis
The tool is being evaluated during this course (2020-2021), using the institutional Moodle server for undergraduate students at our University of Cienfuegos (moodle versions 3.7 and 3.11 latest as of the date of this publication). In a first phase, it is being tested by 5 teachers (3 from the Computer Science area and 2 from Social Sciences) checking the tool with their own courses and reporting errors or proposing improvements.
In addition, it was used for the analysis of the pre-congress courses of the 3rd Scientific Conference of the University of Cienfuegos (cursos.ucf.edu.cu) with 17 courses, with a real data load. The results obtained were highly accepted and useful for teachers and conference coordinators.
In this paper, the Moodle Offline Analytics web application is presented as a basic and simple solution to analyse the records of Moodle courses and provide useful interactive graphics to describe the activity of the students, with great acceptance by managers and teachers who tested the first version available in the national network of Cuban Universities (moodlestats.ucf.edu.cu).
Ahmed, S.A.; Khan, S.I. A Machine Learning Approach to Predict the Engineering Students at Risk of Dropout and Factors Behind: Bangladesh Perspective. In Proceedings of the 10th International Conference on Computing, Communication and Networking Technologies (ICCCNT), Kanpur, India, 6–8 July 2019; pp. 1–6.
Ji, Y.P., Marticorena, R., Pardo, C., López, C., Juez, M. (2020). Monitorización de la actividad y rendimiento de los alumnos en Moodle para su análisis visual. Actas de las Jenui, vol. 5. Páginas: 261-268
Luna, J.M.; Castro, C.; Romero, C. (2017). MDM tool: A data mining framework integrated into Moodle. Comput. Appl. Eng. Educ, 25, 90–102.
Sáiz-Manzanares, M.C.; Rodríguez-Díez, J.J.; Díez-Pastor, J.F.; Rodríguez-Arribas, S.; Marticorena-Sánchez, R.; Ji, Y.P. (2021). Monitoring of Student Learning in Learning Management Systems: An Application of Educational Data Mining Techniques. Appl. Sci., 11, 2677. https://doi.org/10.3390/ app11062677