Machine learning predictions for regioselectivity of hydroformylation reactions: leveraging limited data for high-precision results DOI
Jiaqi Wu, Ye Liu, Xue‐Qing Gong

и другие.

Pure and Applied Chemistry, Год журнала: 2025, Номер unknown

Опубликована: Май 22, 2025

Abstract The development of robust machine learning models to assist the prediction and optimization homogeneously catalyzed reactions has attracted wide interests. In this work, we propose a workflow estimate linear branched ratio products in hydroformylation using stacking ensemble method that integrates Random Forest, eXtreme Gradient Boosting, Light Boosting Machine algorithms, leveraging physicochemically significant features from small-batch experimental data. model achieves superior performance with R 2 Root Mean Square Error values 0.918 0.078, respectively. Moreover, SHapley Additive exPlanations analysis density functional theory calculations reveal impact gap between highest occupied molecular orbital lowest unoccupied alkenes on regioselectivity reactions, indicating larger tend result higher proportion products. This study illustrates combination interpretable can serve as useful strategy for predicting reactions.

Язык: Английский

Machine learning predictions for regioselectivity of hydroformylation reactions: leveraging limited data for high-precision results DOI
Jiaqi Wu, Ye Liu, Xue‐Qing Gong

и другие.

Pure and Applied Chemistry, Год журнала: 2025, Номер unknown

Опубликована: Май 22, 2025

Abstract The development of robust machine learning models to assist the prediction and optimization homogeneously catalyzed reactions has attracted wide interests. In this work, we propose a workflow estimate linear branched ratio products in hydroformylation using stacking ensemble method that integrates Random Forest, eXtreme Gradient Boosting, Light Boosting Machine algorithms, leveraging physicochemically significant features from small-batch experimental data. model achieves superior performance with R 2 Root Mean Square Error values 0.918 0.078, respectively. Moreover, SHapley Additive exPlanations analysis density functional theory calculations reveal impact gap between highest occupied molecular orbital lowest unoccupied alkenes on regioselectivity reactions, indicating larger tend result higher proportion products. This study illustrates combination interpretable can serve as useful strategy for predicting reactions.

Язык: Английский

Процитировано

0