Fig. 3

Variable importance ranked by the Boruta algorithm. The Boruta algorithm identifies relevant features by comparing their importance to that of randomized shadow variables, with higher scores indicating greater contribution to the classification model. The boxplots display the distribution of importance scores for each variable across multiple iterations of the Boruta feature selection process. Variables on the right show higher importance, indicating a stronger contribution to the model’s predictive performance. Shadow variables (in blue) serve as references for identifying truly relevant attributes. Variables with importance scores significantly higher than shadow variables are considered important.