Stable isotope and trace element analyses with non-linear machine learning data analysis improved coffee origin classification and marker selection.

The potential of advanced machine learning models for improvements to origin classification and the identification of relevant origin markers were demonstrated. The decision tree-based models are superior with their embedded variable identification feature and visual interpretation. This article is protected by copyright. All rights reserved.