Improvements to the Apriori Algorithm and Its Application in Educational Decision-Making Systems

Abstract

The Apriori algorithm is among the traditional algorithms that are mainly employed to mine the educational data. Regrettably, the Apriori algorithm is not only the most classical algorithm, but also has some weaknesses, for instance, it has high computational complexity, generates too many candidate itemsets, which in turn leads it to be mainly used in small-scale educational data scenarios only. In this paper, the authors propose a modified version of the Apriori algorithm which aims to the rule generation process and the computational operation more efficient through data preprocessing phase information improvement and using the dynamic thresholding adjusting component as well as a not-so-well-known intelligent technique. The usefulness of the improved algorithm is tested by an experiment with actual course performance assessments, and the algorithm is also applied in the educational decision support system which comprises the course recommendation, teaching strategy optimization, and academic risk prediction the students. The results of the study suggest that there are no gains in script execution speed, generalization rate, or memory usage while there are with great advantage in the run time of the proposed algorithm. Consequently, the proposed algorithm is superior to the traditional one as it provides higher efficiency and thus, could be the best choice for educational data mining.