Predicting Student’s Academic Performance Using Data Mining Methods: Review Paper

The purpose of this study is to identify the most prominent methods used in data mining by researchers to improve student performance in educational institutions. Several papers were reviewed using the descriptive approach. The study revealed that there are new factors affecting student performance, and they differ from the factors in previous studies. The information revolution, the technological era, and the development of the educational system are the cause of the appearance of those new factors. Most papers adopted Random Forest and Naive Bayes as data mining techniques, and through the analysis, it was noted that the vast majority of the studies reviewed employed decision tree algorithms and its variants. The Hybrid approach LMT achieved the highest accuracy percentage among the reviewed papers, taking into consideration the differences in the field of application of these tools, such as study sample, dataset size, study implementation year, and study constraints. A proposed model has been addressed for further research. The study recommends that the Arab countries in general and the Arabian Gulf region should promote the value of adopting these algorithms. This is important to conduct this prediction to Bahrain students to identify the levels of students before the final exam, which is in alignment with the new framework for assessment recently approved for all academic levels.