Complexity

Review Article

A Systematic Review of Deep Learning Approaches to Educational Data Mining

Table 3

Deep Learning approaches in the EDM field: architectures employed, baseline methods, and evaluation measures. The column Performance indicates whether the approaches outperformed baseline methods (>), underperformed (<), or obtained similar results (=).


Reference	Architecture	Baseline	Evaluation	Performance

Abhinav et al., 2018 [38]	MLP	SVD, Slope One, K-NN	MAE, RMSE	>

Akram et al., 2018 [44]	LSTM	Majority class, RF, SVM	Accuracy, Precision, Recall, F1	>

Alam et al., 2018 [25]	MLP	DT, RF, SVM, KNN	Accuracy	>

Alvarado et al., 2018 [43]	WE	N-grams	Precision, Recall, F-measure	=

Aung et al., 2018 [36]	CNN, VGG16, AlexNet	Random guess, LinReg	AUC	>

Bendangnuksung and P., 2018 [20]	FNN	DT, NB, MLP	Accuracy	>

Choi et al., 2017 [45]	WE	-	Krippendorff’s alpha	−

Fei and Yeung, 2015 [28]	RNN, LSTM	SVM, LogReg, IOHMM	AUC	>

Guo et al., 2015 [23]	Autoencoder	NB, SVM, MLP	Accuracy	>

Khajah et al., 2016 [17]	LSTM	BKT	AUC	=

Kim et al., 2018 [27]	BLSTM	LogReg	AUC	>

Kim et al., 2018 [26]	BLSTM	LogReg	AUC	>

Lalwani and Agrawal, 2017 [14]	LSTM	PFA, BKT	AUC	=

Lin and Chi, 2017 [11]	RNN, LSTM	Majority voting, BKT	Accuracy, Precision, Recall, F-measure	>

Mao et al., 2018 [15]	LSTM	BKT, IBKT	RMSE, Accuracy, Recall, F-measure, AUC	=

Min et al., 2016 [33]	LSTM	CRF	Accuracy	=

Montero et al., 2018 [13]	LSTM	BKT	AUC	>

Okubo et al., 2017 [24]	LSTM	LinReg	Accuracy	>

Piech et al., 2015 [10]	RNN, LSTM	BKT	AUC	>

Sales et al., 2018 [46]	LSTM	-	-	−

Sharada et al., 2018 [22]	MLP	RF, LogReg	Log Loss, RMSE, , AUC, Gini, MPCE	<

Sharma et al., 2016 [34]	CNN, AlexNet, VGG16, LSTM	SVM, HMM	Accuracy	>

Taghipour and Ng, 2016 [41]	LSTM	SVR, BLRR	QWK	>

Tang et al., 2016 [21]	LSTM	Majority class	Accuracy	=

Tato et al., 2017 [37]	CNN	SVM, NB, LSA, LDA, MLP	Accuracy, F-measure	>

Teruel and Alemany, 2018 [29]	LSTM	LSTM	AUC, RMSE,	−

Wang et al., 2017 [30]	CNN, RNN	SVM, LogReg, DT, AdaBoost, GTB, RF, GNB	Precision, Recall, F-measure, AUC	=

Wang et al., 2017 [12]	LSTM	LogReg	Accuracy	>

Wang et al., 2017 [53]	LSTM	LogReg	Recall, Precission, F-measure	=

Whitehill et al., 2017 [31]	FNN	-	AUC	−

Wilson et al., 2016 [50]	RNN	IRT, TIRT, HIRT	Accuracy, AUC	<

Wilson et al., 2016 [16]	LSTM	-	-	−

Wong, 2018 [39]	LSTM	-	-	−

Xiong et al., 2016 [18]	LSTM	PFA, BKT	AUC,	>

Xing and Du, 2018 [32]	RNN	DT, KNN, SVM	Accuracy, AUC	>

Yang et al., 2018 [35]	LSTM	-	MSE	−

Yeung and Yeung, 2018 [19]	LSTM	-	AUC	−

Zhang et al., 2016 [42]	DBN	NB, LogReg, DT, Perceptron, SVM	Accuracy, AUC, Precission, Recall, F-measure	>

Zhang et al., 2017 [49]	LSTM, Autoencoder	LSTM	AUC,	>

Zhao et al., 2017 [40]	MN	FNN, SVR, BLRR, LSTM	QWK	>