The pursuit of cleaner and more sustainable fuel alternatives has intensified interest in biodiesel production. Unlike previous studies that primarily focus on reaction conditions, this work introduces a data-driven material–process co-design framework that integrates catalyst precursor selection, preparation conditions, and reaction parameters. A comprehensive dataset made up of 448 data points was compiled from peer-reviewed studies and was utilized to train two machine learning (ML) algorithms, extreme gradient boosting (XGB) and random forest (RF). The input space integrated catalyst precursor type, calcination temperature, and time, textural properties, and transesterification reaction inputs such as reaction time, reaction temperature, catalyst concentration, and methanol-to-oil ratio. Three modeling scenarios were assessed: Case 1 (all inputs), Case 2 (exclusion of catalyst preparation conditions), and Case 3 (exclusion of catalyst properties). Model performance assessment showed that XGB consistently outperformed RF, achieving high R 2 values (0.8829 - 0.9310), low root mean square errors (2.1770 - 2.8357), and lower Akaike information criterion (712.58 - 949.96). Shapley Additive Explanations (SHAP) analysis revealed that the catalyst precursor emerged as the most influential input. Genetic algorithm–based optimization identified an eggshell-derived catalyst prepared at a calcination temperature and time of 897.65 °C and 3.99 h, respectively, with a pore diameter, pore volume, and a surface area of 11.99 nm, 0.19 cm 3 /g, and 45.33 m 2 /g, respectively. The reaction conditions were 64.87 °C, 2.88 h, 3.99 wt%, and 12:1, yielding a predicted biodiesel yield of 99.12%. These findings emphasize the importance of intelligent catalyst precursor selection and preparation in maximizing biodiesel yield. • GA optimized biodiesel yield reached 99.12% under integrated material–process design. • Catalyst precursor–process interplay governs biodiesel production performance. • Eggshell-derived catalyst achieved optimal mesoporosity and surface activity. • SHAP analysis revealed the mechanistic influence of catalyst and process parameters. • Data-driven optimization enables rational catalyst design for sustainable biodiesel.
Amenaghawon et al. (Fri,) studied this question.