Abstract: In response to issues such as strong subjectivity and static data utilization in the formative assessment system for automotive theory courses, this study examines the process assessment data and course grades of 473 student groups from 2020 to 2024. Using a cross-period evaluation method that combines the diffusion model with a multiple linear regression model, the importance of various process assessment items is clarified. Teachers should strengthen the evaluation of classroom performance during formative assessment, implement tiered assessment for in-class experiments, and design stage-specific test content based on competency dimensions to promptly identify learning issues among students.