Data-driven analysis in the selective oligomerization of long-chain linear alpha olefin on zeolite catalysts: A machine learning-based parameter study

In this study, the oligomerization of 1-octene was investigated using various zeolites through both experimental and machine learning (ML) approaches. The structural characteristics of the zeolites and experimental conditions were used as input parameters to analyze the feature importance of each key factor on the oligomerization of 1-octene. By quantifying these influences, the reaction mechanism was elucidated, and a methodology for maximizing the oligomerization reaction was developed. The ML methods employed in this study were SHapley Additive exPlanations (SHAP) and Sure Independence Screening and Sparsifying Operator (SISSO). While the SHAP method is a well-validated conventional approach, it has limitations due to its high data requirements. Therefore, the SISSO method was applied, as it requires fewer data points and offers a transparent computational process. SISSO provided results in the form of human-interpretable equations, allowing for an analysis of these equations to deepen insights into the reaction mechanism. • Linear alpha olefin oligomerization over heterogeneous zeolite catalysts. • Uncover the relationship between physicochemical properties of zeolites and oligomerization. • Development of methodologies combining experimental and machine learning outputs. • Quantitative elucidation of parameter effectiveness in 1-octene oligomerization on zeolites.