Enhancing academic performance prediction with temporal graph networks for massive open online courses

Journal of Big Data

Table 3 Comparisons between the baseline models and APP-TGN in early predicting at-risk students in terms of ACC(%)

Tasks	Duration	ProbSAP	CNN-LSTM	OMLP	HRNs	MTGNN	AP-GT	APP-TGN1	APP-TGN
Pass/Fail	Week5	69.89	71.42	69.13	72.31	72.88	72.81	75.10	77.12*
	Week10	70.81	73.31	70.26	73.22	73.25	73.52	75.89	78.76*
	Week15	72.09	74.31	71.66	74.23	74.64	75.06	77.58	80.05*
	Week20	72.89	76.10	72.11	75.83	76.33	76.91	79.11	81.65*
Pass/Withdrawn	Week5	62.88	63.48	61.92	65.72	64.98	65.06	68.88	72.11*
	Week10	63.37	64.82	62.11	66.30	65.75	66.13	69.86	73.35*
	Week15	65.53	68.09	63.28	67.86	67.04	67.58	70.89	74.43*
	Week20	66.18	69.17	64.26	68.84	69.56	70.01	72.08	75.13*

An asterisk (\(*\)) denotes improvements that are statistically significant when compared to the best performing baseline. These improvements are validated by a two-sided t-test with a p-value less than \(10^{-3}\), confirming their statistical significance