Educational Data Mining with Learning Analytics and Unsupervised Algorithms: Analysis and Diagnosis in Basic Education

This article presents an experience report of an intervention project in Basic Education in a public school, addressing the use of Data Mining and Learning Analytics for teaching and learning processes. In particular, the project aims to identify, among eight mathematical contents, which are those that students have more difficulties. In this context, digital games referring to the subjects were developed, thus raising relevant information for the construction of a dataset, in which Learning Analysis techniques, Educational Data Mining and Unsupervised Learning strategies were used together. Furthermore, for each technique mentioned, different software were applied, such as Jclic and Orange Data Mining Canvas, as well as the Knowledge Discovery in Databases methodology. The results obtained show that it was possible to identify the contents that the students had difficulties through the set of applied approaches, reaching the research goal. The findings reinforce the potential for collecting, using and analyzing data for a more personalized and immersive learning, aiming to conduct and provide assertive feedbacks for teachers and school managers to assist in intelligent decision-making in the educational environment.