Machine learning predictions of optical properties in metal-organic complexes: linear and nonlinear constants

Metal-organic complexes are highly efficient and widely applicable in many optical applications. The efficiency depends on the optical constants. The determination of such constants through experimentation or computational methods is time-consuming. In this work, the ML models were built to predict the linear and nonlinear optical constants of metal-organic complexes. The constants were predicted using various models. The best-performing model was optimized. The models were then analyzed using explainable artificial intelligence (XAI) techniques, and the prediction mechanism was explained. The LSBoost model was found to have better performance. This model was integrated into a material screening tool for NLO applications