Data mining technology can extract important information from numerical data to help students develop better, and a good mental state can help improve students’ academic performance. A decision tree is a machine learning technique that can be utilized to complete data mining tasks and predict students’ mental health. Traditional decision trees can complete this task, but they pay less attention to the monotonic constraint relationship between students’ personal reasons and external environment and the degree of psychological stress. To address this problem, this paper proposes a Kendall’s rank correlation coefficient-based monotonic decision tree model (KCC-tree). This algorithm can consider monotonic constraints in the data mining process, which accurately predicts the stress level of students. With the use of Kendall’s rank correlation coefficient, the proposed KCC-tree method can be used to evaluate the monotonic consistency of features and labels. This approach allows for the selection of qualified candidate features, which are then used to determine the split point. Moreover, the analysis of classification accuracy, mean absolute error and precision reveals that the monotonic classification decision tree method discussed here is feasible for predicting the stress level of students. In addition, this paper illustrates the effectiveness and feasibility of the algorithm by comparing the KCC-tree with classic decision trees such as BFT, C4.5, LAD, and SC.