This dataset was employed in the manuscript:  [1] Löfgren J, Tarasov D, Koitto T, Rinke P, Balakshin M, Todorović M. Machine Learning Optimization of Lignin Properties in Green Biorefineries. ChemRxiv. Cambridge: Cambridge Open Engage; 2022. This dataset contains the results of optimizing the green AquaSolv Omni biorefinery concept using the Bayesian optimization code BOSS. The input data represents the experimental processing conditions for each sample, namely the reactor temperature and P-factor (reaction severity). For each sample, the lignin yield was measured and the number of ß-O-4 linkages, ratio of syringyl and guaiacyl units, and total carbohydrate content were quantified using a 2D NMR HSQC method.  The lignin yield is given in weight percent of the original biomass and the NMR-quantified moieties are measured in terms of number of units per 100 aromatic (Ar) units. The data found in this record comes in two different versions: one for the Combined Acquisitions (CA) and one for the Pure Acquisitions (PA) strategy described in [1]. For each strategy, there is one data file for the training data and one for the test data. For additional details on the experimental processing and Bayesian optimization, please refer to [1].

2022