The Prediction of Student Grades Using Collaborative Filtering in a Course Recommender System

Systematic Review of Recommendation Systems for Course Selection

“…Furthermore, many studies mentioned the use of some evaluation metrics and did not include the exact performance results for some and/or any of these metrics. On the other hand, some papers, such as [32,38], included arguably, too many metrics for evaluation on different datasets. Finally, some publications did not discuss the implementation of the proposed algorithm in enough detail to allow reproduction of the results.…”

“…They list many obstacles to using the current CF models to create a course recommendation engine, such as the absence of ratings and metadata, the uneven distribution of course registrations, and the requirement for course dependency modeling, (4) The system suggested by Malhorta et al [29] will assist students in enrolling in the finest optional courses according to their areas of interest. This method groups students into clusters according to their areas of interest, then utilizes the matrix factorization approach to analyze past performance data of students in those areas to forecast the courses that a specific student in the cluster can enroll in, (5) Authors in [30] proposed the CUDCF (Cross-User-Domain Collaborative Filtering) algorithm, which uses the course score distribution of the most comparable senior students to precisely estimate each student's score in the optional courses, (6) The main aim of the authors in [31] was to improve the precision and recall rate of recommendation results by improving the collaborative filtering algorithm, (7) Students' grade prediction using user-based collaborative filtering was introduced by authors in [32], (8) Authors in [9] improved association rule generation and coverage by clustering, (9) Authors in [33] suggested utilizing big data recommendations in education. According to the student's grades in other topics, this study uses collaborative filtering-based recommendation approaches to suggest elective courses to them, (10) To forecast sophomores' elective course scores, authors in [34] presented the Constrained Matrix Factorization (ConMF) algorithm, which can not only assist students in choosing the appropriate courses but also make the most efficient use of the scarce teaching resources available at universities, (11) Authors in [35] applied the interestingness measure threshold and association rule of data-mining technology to the course recommendation system, (12) Authors in [36] improved the accuracy of recommendations by using the improved cosine similarity, (13) Neural Collaborative Filtering (NCF), a deep learning-based recommender system approach, was presented by authors in [37] to make grade predictions for students enrolled in upcoming courses.…”

Prediction of Students’ Grade by Combining Educational Knowledge Graph and Collaborative Filtering

“…In general, existing student grade prediction methods can be categorized into two main types: one that utilizes regression or classification methods for grade prediction and another that treats the prediction of student grades as a rating prediction problem in recommendation systems. As student learning behavior exhibits time-series characteristics, and recommendation systems' prediction methods rely on users' historical behavioral data, using an analogy to recommendation systems [13][14] to predict student grades yields higher prediction accuracy and stronger interpretability compared to other statistical methods. However, when using recommendation system-based approaches for grade prediction, there is an excessive reliance on students' historical grades, and the accuracy of grade prediction can be affected by courses with low relevance in the historical data.…”

Ranking and filtering of neuropathology features in the machine learning evaluation of dementia studies

“…Previous studies have employed filter methods to identify features related to AD [ 10 , 11 , 12 ]. Gómez‐Ramírez et al focused on self‐reported data, investigating demographics and other relevant factors associated with the development of dementia from mild cognitive impairment (MCI).…”