Fig. 1: Results after applying our XGBoost model to classify arc vs. non-arc basalts.
From: The sub-arc mantle has remained oxidized since the Neoproterozoic oxygenation event
a The relative importance of each input feature in training the model (see “Data Availability”). All feature importances are <0.09, indicating that no single feature dominates the classification model. The color scale shows the relative feature importance. b The confusion matrix comparing known and predicted data labels when the model was applied to the testing subset (30% of the training dataset). The confusion matrix shows a high accuracy of 0.95 and an F1 score of 0.94, suggesting the strong performance of our trained high-dimensional model.
