Resumo:
The study of Machine Learning (ML) has proven to be a significant challenge for beginners, who often face difficulties when starting their studies in this field. This is because mastering it typically requires an understanding of complex algorithms and processes, as well as programming skills. Despite these difficulties, ML technology has become increasingly prevalent in everyday devices and services, making it essential to prepare citizens to become responsible users and creators of intelligent solutions. Given this, there is a growing need to democratize the study of ML, especially among students who rarely have access to this content, such as those in undergraduate programs not closely related to mathematics or computer science. This research thus proposes developing a web-based visual tool, called Visual-AM, to make ML studies more accessible via web browsers, offering an intuitive visual interface to guide beginners in the field. Developed based on the DSRM (Design Science Research Methodology), the tool aims to facilitate a smooth transition between theory and practice through a graphical interface that allows users to work with a variety of ML algorithms using visual elements such as buttons, checkboxes, and sliders. The goal is to provide an interactive experience in which users can manipulate AM models, visualize the effects of their choices in real time, and compare different approaches. The content options offered by the tool include theoretical introductions to AM, practical exercises in classification, regression, clustering, and dimensionality reduction, and quizzes to test users’ knowledge. Visual-AM was evaluated by volunteer undergraduate students enrolled in the Bioprocess Engineering program at UNIFEI in an empirical study using mixed-methods questionnaires. The results indicated excellent acceptance and effectiveness as a teaching aid for the study of AM, across the constructs “Ease of Use,” “Information Flow,” “Engagement,” “Sense of Progress,” and “Contribution of Visual Resources.”
