Correlation between course grades and evaluation grades based on Fruchterman-Reingold and Theil-Sen

To address issues such as fuzzy topological structures, overlooked group differences, and the disconnect between visualisation and quantitative analysis in course grade-teaching evaluation correlation studies, this study proposes an integrated model based on Fruchterman-Reingold and Theil-Sen.Its core innovation lies in constructing a dual-module collaborative architecture: enhancing course community identification through spectrum-guided layout optimisation, and employing topology-featureweighted group regression that integrates topological stability indices with subgroup trend medians to precisely characterise heterogeneous group associations.It implements a closed-loop analytical paradigm of 'topological feature extraction group difference modelling feedback optimisation', overcoming the limitations of linear processes that separate network layout from regression validation.Experimental results demonstrate a convergence efficiency of 0.77%/iteration, outlier robustness of 0.89, and processing time of 87.2 ms.The model achieved a correlation estimation bias of 0.10, group difference identification accuracy of 0.94, and cross-discipline generalisation error of 0.10.In loosely structured course groups, performance declined notably.This model significantly enhances the analytical capability for curriculum interrelationships and improves the accuracy of cross-group correlation estimation in educational assessment, providing reliable technical support for dynamic monitoring of teaching quality and optimisation of interdisciplinary curriculum systems.