Using XgBoost and multimodal physiological signals to identify flight difficulty

Abstract

The focus of this research is to evaluate the performance of XGBoost, a remarkable machine learning algorithm, coupled with SHAP (SHapley Additive exPlanations) for explicability, in interpreting flight difficulty through a spectrum of physiological signals. XGBoost's predictive proficiency is exhaustively assessed against long-established methodologies such as Support Vector Machine (SVM), Regression, and KNN algorithms. The centerpiece of this study is to illuminate the exceptional performance and unprecedented interpretability of XGBoost in conjunction with SHAP, thereby distinguishing it from its conventional counterparts.