Exploring Weka and Python for Educational Data Mining: Naïve Bayes vs. J48

Data mining in the educational field has been commonly used to analyze patterns in academic databases that are hard to analyze manually. The main objective of this paper is to compare the performance of data mining algorithms to predict students’ academic performance. The dataset has been prepared based on students’ academic results, which have been collected from Bangladesh Army International University of Science and Technology. Two data mining algorithms, i.e., Naïve Bayes Classifier and Decision Tree J48 have been implemented, and the resultant accuracy has been compared with each other. After preprocessing (cleaning, discretizing), both the Naïve Bayes Classifier and the Decision Tree J48 algorithm provide accuracy above 88.75%. Here, Weka is also used to compare the accuracy of the raw coding algorithm and the built-in algorithm of Weka. Finally, it is observable that raw coding produces better performance for both algorithms.