Higher Education Students Profiling Analysis Using Decision Tree Algorithm

Grade Point Average (GPA) is one of the indicators used to observe college students’ academic performance in a semester. Currently, it is not known what factors can explain the high or low GPA scores of the students, so it is necessary to carry out students’ profiling. There was a need to explore the correlation between GPA, the courses students took, students’ backgrounds, and the students’ parents’ backgrounds. This research objective was to determine the students’ academic profiling in an observed study program of a university. The student profiling was conducted for the firstyear students. The observed study program needs student profiling to encourage strategy for academic success in the fouryear undergraduate curriculum. The current condition was that the students mostly graduated more than four years in this study program. The students graduated in four years prevalence was 40%. The decision tree algorithm was used for data analysis and students profiling. The data of first-year students was compiled into 550 students. The result of the data analysis formed eight decision trees, five models were statistically significant and three were not. The five models performed well for student profiling in first-year academic students with high and medium-low GPAs. The three models profiling students with medium and low GPAs did not perform well. However, the accuracy of eight decision trees were high, the lower accuracy model was 76.47% and the higher was 90.38%.