A team from the University of Córdoba has designed a model based on fuzzy logic that predicts the performance of online education students, organizing it into four categories and helping teachers give more personalized responses to each student according to their situation.
Distance education has democratized access to knowledge, eliminating problems of time and space. The flexibility and accessibility of this type of systems has increased the number of people who train through online platforms. However, due to the large number of students and the lack of close interaction that the classroom allows, teachers face a big problem: the difficulty of monitoring and adapting learning to the students.
Tools based on artificial intelligence can be allies of teachers, helping to predict student performance so that they can adapt educational strategies to their learning situation.
To facilitate this adaptation and improve online education, a team from the University of Córdoba made up of researchers Juan Carlos Gámez, Aurora Esteban, Francisco Javier Rodríguez and Amelia Zafra have developed an algorithm that predicts student performance in four classifications. different.
Compared to other previous models that predict this performance only from the perspective of “pass or fail” or “leave or continue” the course, “this algorithm based on ordinal classification and fuzzy logic allows predicting the performance of the students while maintaining the order relationships between the categories: abandonment, failure, passed and distinction” explains the researcher from the Department of Computer Science and Numerical Analysis of the UCO, Amelia Zafra.
In this way, the algorithm FlexNSLVOrd It makes a better prediction, but it also allows teachers to better adapt their strategies depending on the classification in which the students find themselves.
The two advantages proposed by this development are the use of ordinal classification with a cost matrix that allows modeling the weight of ordinal classes in learning and allows this ranking to be made. more specific, and the adapted fuzzy logic that, as the researcher from the Department of Electronic and Computer Engineering of the UCO, Juan Carlos Gámez, points out “allows you a certain flexibility since, on the one hand, compared to the standard logic that works with values “Concretely, fuzzy logic works with a range of values, and on the other hand, it automatically adapts to the problem using reasoning closer to what we do in our daily lives.”
The model is fed by the data that the online teaching system generates. That is, the characteristics that it takes into account to predict performance are, for example, the completion of specific tasks and questionnaires, their grade and the clicks that students make on the different resources available on the platform.
For researchers, “interpretability” is also notable, that is, the possibility of understanding the results it produces. And the thing is that, after tracking this behavior of the students, the model performs the classification, but it also makes itself understood, since “in contrast to the black box algorithms that tell you if the student will pass or drop out, but they do not tell you how or why”, this new tool provides a series of rules for each category that show the most relevant resources and activities that the student must carry out”, continues Zafra.
Thus, the algorithm could help teachers identify students and be able to use reinforcements or strategies that “for example, rescue students with problems.”
In fact, in this sense, the algorithm allows teachers to even know what type of characteristics are decisive or not to know performance. “Perhaps a task that the professor thought was important for him turns out to be not decisive in knowing whether he will finally pass or fail,” explains Francisco Javier Rodríguez, also a researcher at the Department of Electronic and Computer Engineering at the UCO.
The algorithm has been tested using a very large set of publicly available Open University Learning Data (OULAD), which is openly available and comprises a large sample of students and courses. The future use of this algorithm could include including it as an application in online education platforms (such as Moodle) and automatically giving feedback on the students’ performance to the teachers.