Stress in students can be caused by various factors. Academic pressure is a significant source, with heavy workloads, tight deadlines, and high expectations from parents and teachers. Social factors, financial concerns and personal issues, exacerbate stress. Detecting and intervening in student stress is crucial for maintaining academic performance, mental and physical health, social relationships, and long-term success. Early intervention helps prevent the negative effects of stress and supports the overall well-being of students. Machine learning helps predict stress levels by analyzing patterns in physiological and behavioral data. This research explored the impact of machine learning algorithms on decoding student stress patterns, focusing particularly on LightGBM model. The performance of LightGBM model is compared against decision tree, random forest, Naïve Bayes, and GBM. The experimental results reveal an accuracy score of 96% for LightGBM. Further research in this domain can focus on developing targeted interventions to mitigate negative impacts of mental stress on students.