Automating Grading to Enhance Student Feedback and Efficiency in Higher Education with a Hybrid Ensemble Learning Model

Feedback utilisation in the classroom is well-documented, with various research evidencing its beneficial effect on students' learning capacity. Nevertheless, the majority of research endorsing feedback frameworks and tactics is derived from studies involving children mandated to attend school, whereas effective feedback in higher education is comparatively under-examined. This systematic review seeks to address this gap by evaluating the efficacy of various feedback mechanisms in educational contexts. The research analyses student feedback in higher education by contrasting three supervised classifiers—SVM, RHEL, and DT—alongside three feature selection techniques: IG, mRMR, and CFS. The model development consists of three essential phases: data preprocessing, feature selection, and model training. A hybrid ensemble learning method utilising ELM and RH network is employed to correlate SF modes with retrieved attributes. Research demonstrates that low-stakes quizzing is a successful approach, although tutor and peer feedback can be advantageous contingent upon implementation variables. Experimental findings indicate that the CFS approach, in conjunction with the RHEL classifier, attained the maximum accuracy of 99.36%. This study underscores the importance of genuine student feedback systems in higher education, illustrating that their proper implementation can enhance learning results.