Abstract

This study presents the application of a Gaussian Process Regression (GPR) model for predicting the maximum absorption wavelength (λmax) of lead(II) complex with organic reagent [3-((2-chloro-4- hydroxyphenyl) diazenyl)-5nitrobenzene-1,2-diol] (CHPDND) extracted using various organic solvents. The proposed model utilizes key solvatochromatic parameters, including dielectric constant (ε), polarity/ polarizability (π*), hydrogen bond donor acidity (α), hydrogen bond acceptor basicity (β), and a custom polarity scale (ξ), to model the complex solvent effects on electronic transitions. The GPR model, equipped with a squared exponential kernel, demonstrated high predictive accuracy and efficiency, capturing the nonlinear relationships between solvent properties and λmax. The model underwent automatic hyperparameter optimization, achieving convergence within minimal iterations and exhibiting stable performance. The findings were validated using both experimental and aggregated datasets, showing a strong agreement between predicted and actual values. The results emphasize the potential of machine learning approaches, particularly GPR, as reliable and efficient alternatives to traditional computational methods for optical property prediction. This work provides a practical framework for solvent selection and molecular design in the fields of analytical chemistry and materials science.

Keywords: solvatochromism, λmax prediction, gaussian process regression, solvent effects, lead complexes

Article language: En.