Predictive Analysis of Engineering Students’ Enrollment in Entrepreneurship Education Courses

Entrepreneurship education (EE) courses have grown in engineering education to prepare graduates to be competitive in today’s innovation economy. While literature provides empirical support for the benefits of EE courses, there is minimal research examining students’ enrollment in EE courses. The present study is conducted with institutional records of 12,045 undergraduate engineering students at a public research university in the United States. The first objective is to evaluate the performance of the decision trees (CART), random forest, and XGBoost algorithms when predicting engineering students’ enrollment in EE courses. The second objective is to examine the variable importance of the engineering students’ academic (undergraduate GPA, high school GPA, SAT scores, major), demographic (sex and underrepresented minority status), and socioeconomic (family income and parents’ educational level) backgrounds in predicting their enrollment in EE courses. The results indicate that ensemble methods, random forest and XGboost, perform better than CART on all the evaluated performance metrics (accuracy, specificity, recall, precision, and the F1 score). Furthermore, the results suggest that variables measuring students’ academic background held the highest predictive importance, followed by the variables pertaining to students’ socioeconomic and demographic backgrounds. Educational practitioners, administrators, and policymakers may use the research results to strategize resources and efforts to initiate or revise entrepreneurship programming for engineering students. The study lays the groundwork for an evaluative examination of the relatively under-utilized machine learning methods in engineering education research, encouraging further methodological inquiry on the topic.