Increasing school dropout rates are a problem in many educational systems, with student disengagement being one significant factor. Learning analytics (LA) is a new field with a key role in educational institutions in the coming years. It may help make strategic decisions to reduce student disengagement. The use of technology in educational environments has grown significantly and, with it, awareness of the importance of student engagement. We exploit tracking and wearable technologies to increase user engagement in learning processes, exploring also the area of multi modal learning analytics (MMLA). We use WIoTED (wearables and Internet of Things for education), an interactive and collaborative system designed to improve motivation and learning. This paper presents the results obtained in different experiments conducted in a secondary school in a long-term participatory learning context. The captured data were analyzed and used to identify different students' behavior patterns, showing their progress and motivation. Subsequently, from the captured data, and aiming at a decision-making phase, we used machine learning techniques and MMLA methodologies to construct models able to 'explain' when student engagement is present, so this knowledge can later be exploited. In particular, we chose decision trees and rule systems based on a set of variables with proven relevance to the problem. The evaluation of this novel engagement classification system confirms the high performance of these variables. The rules obtained, which can be easily interpreted by a non-expert, help the teacher to observe, analyze and make decisions with the purpose of fostering engagement.

This paper presents a web application that retrieves songs from YouTube and classifies them into music genres. The tool explained in this study is based on models trained using the musical collection data from Audioset. For this purpose, we have used classifiers from distinct Machine Learning paradigms: Probabilistic Graphical Models (Naive Bayes), Feed-forward and Recurrent Neural Networks and Support Vector Machines (SVMs). All these models were trained in a multi-label classification scenario. Because genres may vary along a song's timeline, we perform classification in chunks of ten seconds. This capability is enabled by Audioset, which offers 10-second samples. The visualization output presents this temporal information in real time, synced with the music video being played, presenting classification results in stacked area charts, where scores for the top-10 labels obtained per chunk are shown. We briefly explain the theoretical and scientific basis of the problem and the proposed classifiers. Subsequently, we show how the application works in practice, using three distinct songs as cases of study, which are then analyzed and compared with online categorizations to discuss models performance and music genre classification challenges.