Academic Analytics: Applying C4.5 Decision Tree Algorithm in Predicting Success in the Licensure Examination of Graduates

Abstract

This study predicted success in The Licensure Examinations for Teachers (LET) applying C4.5 decision tree algorithm. This system was eyed to help the students and the academia improve their success rates in the LET. The following were the academic areas considered: Entrance examination raw scores; general weighted average (GWA) in the major field of specialization and in professional education and general education subjects; and, whether pass or fail in LET review results as well as LET Board results. There were a total of 348 instances studied, spread over a 5-year period, from 2012 to 2017. The size of the C4.5 pruned tree totaled to 16 with 10 leaves. The predictive capacity of the system was found to be almost perfect, evidenced by the generated Kappa value of 0.8195. Therefore, should there be any predicted failure, the teachers and deans could promptly provide intervention programs to improve the scores and GWA of students, which would eventually redound to the success in the LET. This study was conducted at Aklan State University (ASU), (Philippines), because at current times, there is still no academic analytics system installed for this purpose. Consequently, ASU was considered as the test case and the datasets utilized in this study were from ASU. In essence, this study could help in accreditation matters, particularly in tracking down the LET success rates of the teacher graduates of HEIs, ASU included.