Machine learning analysis of fatty acid composition in washingtonia filifera and sabal palmetto palm kernels for potential industrial applications

Abstract

The fruits of many palm species, including Washingtonia filifera (W. filifera) and Sabal palmetto (S. palmetto), are rich in secondary metabolites and widely used for food and industrial purposes. This study aimed to analyze the fatty acid profiles of the kernels of these two underutilized palm species and to develop predictive models using machine learning to assess their potential for diverse industrial applications. Oleic acid was the most abundant fatty acid, constituting 37.13% in W. filifera and 33.29% in S. palmetto. Lauric acid followed at 25.80% in W. filifera and 25.57% in S. palmetto. Linoleic and myristic acids were also prevalent, with varying ranks between the two species. Total fat, total Saturated Fatty Acids (ΣSFA), total Monounsaturated Fatty Acids (ΣMUFA), and total Polyunsaturated Fatty Acids (ΣPUFA) were analyzed using machine learning (ML) models. The performances of Support Vector Machine (SVM), Random Forest (RF), Extreme Gradient Boosting (XGBoost), and Multilayer Perceptron (MLP) models were evaluated using metrics like RMSE, R² score, and MAE. SVM achieved the highest R² scores (0.98-0.99), demonstrating its effectiveness in accurately predicting fatty acid profiles, which is crucial for assessing their suitability for various industrial uses, including food, biodiesel and potential applications in the cosmetic industry.