A preliminary implementation of data mining approaches for predicting the results of English exit exam

One of the key roles of Loei Rajabhat University in student services is to improve their students' English language proficiency. This kind of skill is accepted as an essential requirement of job recruitment. In this paper, we attempt to primarily analyze the results of English tests and student data by using data mining approaches in order to explore determinants of the results of English exit exams. The research applied multiple data mining approaches namely Naïve Bayes, Bayesian network (Bayes net), decision tree (C4.5) and SVM, to the dataset of graduating students in 2015. By comparing their accuracy, C4.5 outperformed other approaches in predicting the results of English exit exam. The analysis also indicates that “the results of English placement test” was the strongest predictor of the results of English exit exam. Therefore, the results of English placement test were analyzed further against the results of English exit exam by using the linear regression. The main implication of this research is that the university should focus more significantly on students who fail the English placement test and the tests can be arranged more frequently. The future work is needed to construct the C4.5 predictive model for further prediction of the exit exam.