Machine Learning of Photocatalytic Reductive Coupling of Aldehydes and Its Interpretation Using SHapley Additive exPlanations

Photocatalytic coupling reactions are valuable for sustainable synthesis, yet their development and optimization still rely on empirical time-consuming screening. Using 439 literature data points, we developed an interpretable machine learning workflow for photocatalytic aldehyde coupling reactions. SHAP analysis identified key steric and electronic descriptors, while external validation supported the model robustness. These data-driven approaches provide practical guidance for reaction understanding and optimization.