Table 1.

Literature review on approaches for academic performance assessment or prediction

References	Method	Contribution	Dataset	# of courses	Deep learning	Machine learning	Data analysis	Prediction range	Metric	Score
Chen et al. [4]	Multi-layer perceptron with evolutionary algorithms, for predictive analysis of academic performance	Comparison between Cuckoo and Gravitational search algorithms	Small	3	✗	✔	✗	Continuous (0–10)	MAE, RMSE, MSE, R	0.72 (R2)
Livieris et al. [10]	Multi-layer perceptron, SVM, machine learning algorithms	A machine learning algorithm interface (tool)	Small	1	✗	✔	✗	Categorical (4 class)	Accuracy	0.86
Li et al. [9]	Fuzzy clustering and linear regression with Random Forest and SVM	Fuzzy c-means clustering implementation	Large	3	✗	✔	✗	Categorical (20 class)	Accuracy	0.79
Al-Shehri et al. [1]	Academic performance prediction using SVM and k-NN	Comparison between machine learning algorithms	Small	2	✗	✔	✗	Categorical (20 class)	MAE, RMSE, MSE, R	0.82 (R2)
Patil et al. [12]	LSTM-based sequential modeling of grades	Comparison of deep and non-deep methods	Large	5	✔	✔	✗	Categorical (7 class)	RMSE, accuracy	0.92
Harvey et al. [7]	Machine learning for feature analysis and prediction in K-12	Predictive study of K-12 education dataset	Large	1	✔	✗	✔	Continuous (0–100)	Accuracy	0.71
Proposed approach	Deep learning methods (LSTM, GRU, VAE) with multiple machine learning algorithms	Integration of generative and temporal deep learning approaches with machine learning algorithms and statistical study of the features of the dataset	Large	15	✔	✔	✔	Continuous (0–100)	MAE, RMSE, MSE, R2	0.94 (R2)

Note that our work covers a large number of courses with much finer prediction range and better R²-score