The foundation of higher vocational education curriculum innovation and reform is how to produce highly competent professionals in accordance with societal needs, in response to the rapid modernization of vocational education. In this study, digital technology is used to support the school’s comprehensive management service platform for information technology instruction. The enhanced k-means algorithm and Aprior algorithm are then used to analyze college students’ behavioral data over the course of their education and identify the rules of correlation between students’ behaviors and academic achievements. The article presents a unique ANN-CBL academic performance prediction model with an attention mechanism. The model’s representation capability is strengthened by using an attention mechanism to assign weighted values to the outputs. The fully connected layer receives the weighted time-series information in order to estimate the students’ academic achievement. The model suggested in this paper is able to identify the characteristics of the behaviors and grades in each semester and achieves better prediction accuracy with good interpretability. The experimental results indicate that by using the optimized k-means clustering algorithm to cluster the students in each index, the clustering outcomes will help the student management workers grasp the status of the students and offer decision support for enhancing the efficiency of educational management classification guidance for the students.
