Prediction of the hydrophobicity of platinum(IV) complexes based on molecular surface properties

A quantitative structure–property relationship (QSPR) study was performed for predicting the hydrophobicity of Pt(IV) complexes. Two four-parameter equations, one based solely on structural descriptors derived from electrostatic potentials (ESPs) molecular surface, and other integrated ESP with surface area (AS), were firstly constructed. Mechanistic interpretations introduced elucidated in terms solute-solvent intermolecular interactions. Subsequently, several up-to-date modeling techniques, including support vector machine (SVM), least-squares (LSSVM), random forest (RF) Gaussian process (GP), utilized to build nonlinear models. Systematical validations leave-one-out cross-validation, validation test set, as well a more rigorous Monte Carlo cross-validation verify reliability constructed The predictive performances four different methods follow order LSSVM≈GP > RF SVM. pure-ESP-based models are generally inferior AS-integrated ones. Comparisons previous results made.