Explainable Feature Ranking Using Interactive Dashboards
Diogo Amorim, Matilde P. M. Pato, Nuno Datia · 2024
In the dynamic realm of machine learning, achieving transparency and understandability is crucial for fostering trust and facilitating broader adoption. This study presents an enhanced version of the Ensemble Feature Ranking algorithm, tailored to optimize feature selection in machine learning models. This paper proposes the use of an interactive dashboard application, as part of learning environment, designed to provide users with a visually intuitive platform for exploring the algorithm's internal metrics and rankings. The dashboard facilitates a deeper understanding of feature importance and algorithm behaviour, bridging the gap between complex algorithms and user comprehension. By combining advanced algorithmic techniques with a user-centric interface, our approach promotes transparency, accountability and increased user engagement in the explanation of machine learning models.