Enhancing machine learning in gas–solid interaction analysis: Addressing feature selection and dimensionality challenges

Abstract

This paper addresses the critical importance of accurate analysis in research, emphasizing the necessity of error-free and unbiased calculations. While ground truth values are pivotal for validating accuracy, their absence poses challenges in feature importance, feature selection, and clustering methods commonly used in machine learning. Liu et al. have introduced innovative models targeting gas-solid interactions, but their reliance on model-specific methodologies raises concerns about potential biases and erroneous conclusions. This study advocates for robust statistical validation techniques, including the application of Variance Inflation Factor (VIF), Spearman's correlation, and Kendall's tau, to enhance the reliability of feature selection and ensure more accurate insights. By emphasizing a rigorous approach to statistical significance, this paper aims to improve the interpretability and effectiveness of machine learning applications in this specialized field.