Insight to the prediction of CO 2 solubility in ionic liquids based on the interpretable machine learning model
Ao Yang,Shirui Sun,Yang Su,Zong Yang Kong,Jingzheng Ren,Weifeng Shen
DOI: https://doi.org/10.1016/j.ces.2024.120266
IF: 4.7
2024-05-23
Chemical Engineering Science
Abstract:In this work, we investigated three different ML-based models, i.e., GPR, LightGBM, and CatBoost, for predicting the solubility of CO 2 in various ionic liquids (ILs). Three molecular descriptors, i.e., group contribution (GC), molecular structure descriptors (MSD), and hybrid GC-MSD are used in our three models. The performance of our developed models were rigorously evaluated using mean absolute error (MAE), coefficient of determination (R 2 ), and mean relative error (MRE) (i.e., relative deviation in percentage), with each model subjected to multiple tests employing different random state parameters. The dataset underwent partitioning into training and testing sets at an 80:20 ratio, with additional splits at various ratios to assess prediction performance sensitivity. Overall, all models exhibited proficient CO 2 solubility prediction in ILs, with performance varying based on descriptor type. Notably, the hybrid GC-MSD consistently outperformed others, attributed to GC-MSD incorporates a broader array of molecular feature information. Particularly, the CatBoost-GC-MSD model excelled, achieving an impressive R 2 of 0.9925, MAE of 0.0122, and MRE of 11.1550%. Comparing our models to previous studies revealed the superior performance of CatBoost-GC-MSD across all descriptor types. Furthermore, our model interpretation, employing SHAP analysis, identified pressure, temperature, Chi0, Kappa2, and EState_VSA10 as the top five influential input features. These findings provide valuable insights into the underlying molecular features affecting CO 2 solubility in ILs and lay the foundation for future research in this field.
engineering, chemical