In guiding accounting undergraduates toward informed choices about their professional futures within higher education institutions, it is necessary to predict students’ career growth paths. In this paper, the clustering algorithm integrated with K-Means, K-Modes and GMM is used to study the career growth path portrait of accounting students. Then the convolutional neural network model optimized by cuckoo algorithm is proposed to predict the career growth path of accounting majors. Four career trajectory profiles emerge from the analysis of accounting students’ growth patterns, each representing a meaningfully distinct developmental path. The prediction accuracy of the CSO-CNN model is improved by 3.84%~8.42% compared with the comparison model, and its overall prediction accuracy of 91.33% for the groups with different career growth path portraits, and the training and testing time is smaller, which reveals the good prediction accuracy of the CSO-CNN model for the students’ career growth path. Students’ career growth paths with good prediction accuracy and prediction efficiency.
