arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2512.04652cs.CY
收藏

量化学术流动:一项工程学院(1980-2019)的网络与聚类分析研究(学位转换、计划变更与重新入学)

Quantised Academic Mobility: Network and Cluster Analysis of Degree Switching, Plan Changes, and Re-entries in an Engineering Faculty (1980-2019)

National University of Tucumán(图库曼国立大学)

H. R. Paz

AI总结:

本研究通过CAPIRE框架分析工程学院学生学术流动,揭示学生轨迹的复杂量化结构,识别出六类学生类型,并发现电子和计算等专业作为系统性吸引点。

摘要:

本研究挑战了传统二元学生流动观念(保留与退学),通过将学术轨迹视为复杂、量化路径来重新概念化。利用来自阿根廷工程学院的40年纵向数据集(N=24,016),我们引入了CAPIRE分析框架,区分了学位主修转换、课程计划变更以及相同计划重新入学。虽然73.3%的学生遵循线性轨迹(Estables),但显著的26.7%表现出复杂的流动模式。通过应用主成分分析(PCA)和DBSCAN聚类,我们揭示这些轨迹并非连续,而是结构上量化,占据离散的复杂性带。分析识别出六个不同的学生类型,包括‘转换者’(10.7%)重新定向职业,以及‘稳定重新入学者’(6.9%)表现出停学行为而不改变学科。进一步,网络分析突显了特定的‘枢纽专业’——如电子和计算——作为系统性吸引点。这些发现表明,学生流动是一种有组织的生态系统特征,而非随机噪声,为机构提供了新的课程分析和预测建模视角。

英文摘要:

This study challenges the traditional binary view of student progression (retention versus dropout) by conceptualising academic trajectories as complex, quantised pathways. Utilising a 40-year longitudinal dataset from an Argentine engineering faculty (N = 24,016), we introduce CAPIRE, an analytical framework that differentiates between degree major switches, curriculum plan changes, and same-plan re-entries. While 73.3 per cent of students follow linear trajectories (Estables), a significant 26.7 per cent exhibit complex mobility patterns. By applying Principal Component Analysis (PCA) and DBSCAN clustering, we reveal that these trajectories are not continuous but structurally quantised, occupying discrete bands of complexity. The analysis identifies six distinct student archetypes, including 'Switchers' (10.7 per cent) who reorient vocationally, and 'Stable Re-entrants' (6.9 per cent) who exhibit stop-out behaviours without changing discipline. Furthermore, network analysis highlights specific 'hub majors' - such as electronics and computing - that act as systemic attractors. These findings suggest that student flux is an organised ecosystemic feature rather than random noise, offering institutions a new lens for curriculum analytics and predictive modelling.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑