Analysis of College Students' Mental Health Based on XGBoost

With the rapid development of social economy and the change of students' growth environment, students' mental health problems have become increasingly prominent. In order to comprehensively and systematically understand the psychological stress of college students. In this paper, data mining technology and machine learning methods are used to intelligently evaluate the mental health of college students. By collecting relevant data and combining with SCL-90 mental health evaluation system, the machine learning algorithm is used to accurately simplify the questionnaire questions of the original evaluation system. And to determine the contribution of different indicators to the psychological stress of college students. Based on the simplified data, a psychological stress evaluation model for college students is established to realize the classification evaluation of mental health status. And provide a specific evaluation score or grade. Improve the accuracy of assessment through model training and optimization, and provide data support for follow-up mental health work. In addition, Kendall's Tau-B correlation analysis and XGBoost are used for feature selection, and key feature factors are retained. The method of this paper is helpful to accurately locate the psychological stress of college students, and provide a clear direction and basis for psychological intervention.