基于偏正态分布的视角：优化高校课程成绩评定方法的探讨
Perspective Based on Skewed Normal Distribution: Exploring the Optimization of Course Grade Evaluation Methods in Higher Education

高校课程成绩评定传统上依赖于正态分布模型，但在处理高分段或低分段的偏斜数据时存在局限性。本文基于偏正态分布，探讨其在学生成绩分析中的优化应用。通过实证分析深圳技术大学四门课程(程序设计基础A、高等数学A、离散数学、面向对象程序设计)的成绩数据，对比正态分布与偏正态分布的拟合效果。结果表明，偏正态分布通过引入偏度参数，能够更精准地捕捉成绩数据的非对称特性，其拟合优度(AIC与BIC值)显著优于正态分布，且置信区间更精确。研究进一步提出改进建议：区分合格性考试与选拔性考试、综合过程性评价与终结性考核、增强评分透明度，并合理利用大数据技术优化教学评估。本文为高校成绩评定提供了新视角，支持教育评价体系的科学化与多元化发展，助力创新人才培养目标的实现。
Traditional assessment of university course grades has relied on the normal distribution model, yet this approach shows limitations when handling skewed data in high or low score segments. This study explores the optimized application of skew-normal distribution in student performance analysis. Through empirical analysis of grade data from four courses at Shenzhen Technology University (Fundamentals of Programming A, Advanced Mathematics A, Discrete Mathematics, and Object-Oriented Programming), we compared the fitting effectiveness between normal distribution and skew-normal distribution. Results demonstrate that the skew-normal distribution, by incorporating a skewness parameter, more accurately captures the asymmetric characteristics of grade data, with significantly better goodness-of-fit (AIC and BIC values) than normal distribution and more precise confidence intervals. The study further proposes improvement strategies: distinguishing qualifying examinations from selective assessments, integrating formative and summative evaluations, enhancing grading transparency, and appropriately utilizing big data technology to optimize teaching evaluation. This research provides new perspectives for academic performance assessment, supporting the scientific and diversified development of educational evaluation systems, thereby facilitating the cultivation of innovative talents.