Abstract

Borylationreactions catalyzed by cobalt and nickel compounds occupytheir important niche in synthetic organic chemistry; however, thesearch of parameters for high-yield reactions can be time-consumingand expensive. Recently, machine learning-based regression modelswere able to accurately predict reactivity yields, still when datafrom the literature are used, less accurate models are obtained. Inthis work, transforming the regression problem into a classificationproblem, we managed to predict high-yield cobalt- and nickel-catalyzedborylations using reaction data taken from the literature. With theRandom Forest algorithm, we achieve to get the area under the receiveroperating characteristics (ROC) curve mean values of 0.93 for cobalt-catalyzedreaction models and 0.86 for nickel-catalyzed reaction models. Inaddition, the feature importance indicates that for Co-catalyzed reactions,the characteristics of the catalyst are the most important, whilein Ni-catalyzed borylations, there is a greater influence of the characteristicsof the reactants and products. We think that this study may be a viablealternative to take advantage of reported reactions and could be especiallyuseful for those laboratories that do not have the possibility toperform high-throughput experimentation to optimize their catalyticreactions.