Novel Predictive\nModels for the Heat Capacity of Deep\nEutectic Solvents Using Coupled Atomic/Group Contributions and Machine\nLearning Methods

Deep eutectic solvents (DESs) are novel green solvents.\nPotential\napplications of DESs require a knowledge of their physical and thermodynamic\nproperties. This study is devoted to the DES heat capacity. Since\nthe potential number of DESs to be prepared in the future is innumerable,\nit is vital to have predictive models. In this study, two machine\nlearning models, namely, the multilayer perceptron artificial neural\nnetwork (MLPANN) and the least square support vector machine (LSSVM)\nwere coupled with the group contribution (GC) and atomic contribution\n(AC) approaches. In the contribution methods, each structural fragment\nof the compounds is considered as input to the machine learning models,\nsignificantly enhancing predictive capability. A comprehensive database\nwas collected, including 640 data points from 51 different DESs at\nvarious temperatures. The MLPANN-GC and LSSVM-GC models resulted in\nAARD% values of 1.74 and 1.73%, respectively, while the corresponding\nvalues were 2.90 and 2.64% for the MLPANN-AC and LSSVM-AC models.