Proceedings of The 3rd International Conference on Advanced Research in Education
Student Success Prediction in International Business
Mees Brinkman, Wout van Velzen, Hani Al-Ers, Cornelis Beyers, Xiao Peng
Approximately half of the students of The International Business program at The Hague University of Applied Sciences (THUAS) found in this program do not successfully complete the propaedeutic phase. There has also been a significant increase in the number of enrolled students. This makes good study advice important. This research aims at improving study advice by accurately predicting if a student will successfully complete the propaedeutic phase.
This paper explores the socio-demographic factors (age, type of the previous education, ethnicity, 1st nationality), previous education grades (high-school Grade Point Average (GPA), high-school math and English grades) and intake test results (intake math and intake English) that may predict a student’s success.
In total, there is data available from 3349 students who joined the International Business program in the years 2010 to 2017, each student having up to 50 different factors. This data has been used to train and test several predictive models: the Classification And Regression Tree (CART) model and the Random forest model, which were separately optimized to find the best combination of factors to predict if a student will successfully complete the propaedeutic phase.
This research concludes that average high-school grades, intake math and intake English grades are exceptionally good predictors. The most accurate model turned out to be the Random forest model with a 61.9% accuracy, similar to the accuracy of the CART model which reached an accuracy of 61.3%, but for some groups of students, a much higher accuracy might be reached.
Keywords: CART; Education; Machine learning; Modelling; Student attrition.