Building Machine Learning Models for Reactivity Prediction in Radiation-Induced Graft Polymerization Using Interpretable Parameters

Machine learning prediction models for radiation-induced graft polymerization reactivity of methacrylate monomers were feasibly built with chemically interpretable parameters. The reactivity can be predicted based on the decision-tree-based machine learning algorithms. Among these algorithms, the XGBoost algorithm exhibited a good performance using five interpretable parameters: the solvation free energy of the methacrylate monomer in water, solvation free energy of the methacrylate monomer in hexane, methacrylate monomer radius, conformational entropy of the methacrylate monomers, and the dipole moments of the methacrylate monomers. The machine learning model building resulted in effective reactivity predictions and unveiled important factors for the radiation-induced graft polymerization in a chemically interpretable fashion.