 In this study, we developed a methodology using laser-induced breakdown spectroscopy, LIBES, combined with machine learning algorithms to rapidly detect and analyze the quality of coal samples. We found that K-means clustering, naive Bayes classification, and partial least squares regression can accurately identify the origin of coal samples based on their spectral characteristics. Additionally, these models can also predict the proximate analysis indicators such as volatile matter and fixed carbon with high accuracy. This research provides a reference for selecting appropriate machine learning algorithms for LIBES applications and coal quality assessment. This article is authored by Yanning Zhen, Qing Meilu, Anqi Chen, and others.