Predicting Student Academic Performance at Higher Education Using Data Mining: A Systematic Review

Table 1

Summary of the related studies on predicting drop out among students.

Ref	Techniques	Results	Study sample size	Findings
[21]	DT	Precision (98%)	12969	Student having a GPA <5.79 is more likely to drop out
[22]	Ensemble	Accuracy (92.18%)	261	The ensemble model improves the prediction performance and solves the overfitting issue
[23]	DT	Accuracy (81.01%)	206	Six features related to the first-semester academic factors and parents’ income are the most influencing factors
[24]	Clustering & BPR	—	561	Low exams grades and motivation level may lead to evasion.
[25]	Ensemble	Accuracy (99%)	499	The previous semester GPA is the most influencing attribute in students’ academic achievement
[26]	RF	Accuracy Model_1 (78.84%)	38,842	The common RF package in R obtained the highest output
		Model_2 (47.41%)
[27]	ANN	Accuracy (85.8%)	10,196	The best predictive model was achieved by considering the data related to the first-semester academic performance
[28]	BN	Accuracy (98.08%)	104	Student’s attendance and GPA are the highest impacted features on student’s performance