A Machine Learning Approach to Examine Personality, Adjustment, and Engineering Identity Among College Engineering Students

Background: This study examined whether and to what extent general personality traits based on the Big Five model (McCrae & Costa, 2003), vocational personality types based on Hollands’ Holland’s Self-Directed Search (1985), and students’ adjustment to college as measured by the Student Adaptation to College Questionnaire (SACQ; Baker & Siryk, 1999) predicted academic achievement and engineering identity among undergraduate engineering students by using machine learning models. Hypotheses: We anticipated that general personality traits based on the Big Five model (McCrae & Costa, 2003), vocational personality types based on Holland (1985), and students’ adjustment to college as measured by the SACQ (Baker & Siryk, 1999) would predict academic achievement and engineering identity among undergraduate engineering students. Design/Methods: We cast predicting cumulative GPA (CGPA) and engineering identity measure (EIM) as two separate binary classification tasks where machine learning models were trained to recognize students who achieved above-average CGPA and EIM scores. Results: For the CGPA, we identified 12 features with above-average contributions to the model output. Of these, all were positive predictors except for SACQ-PEA, NEO-Extraversion, SACQ-ATT, and NEO-Neuroticism. For the EIM, we identified 11 features with above-average contributions to the model output, all of which were positive predictors. Neo-neuroticism was a negative predictor, and its contribution was slightly below average (i.e., ranked 12th). Conclusion: Personality factors and college adjustment play an important role in college engineering students’ academic performance and engineering identity. Future researchers should examine factors affecting academic achievement, engineering identity, and adjustment to college among students from diverse backgrounds.