A Comparative Analysis of Machine Learning Techniques for Predicting Student Persistence in Computing Programs
DescriptionJobs in computing fields (consisting of students in computer science, computer engineering, and information sciences) are expected to flourish within the next ten years. In particular, growth will increase 32% for information security analysts and 26% for software developers. Despite the professional need for more graduates, undergraduate students in computer or information science have a 59% rate of attrition. Since pathway patterns for computing students are different from engineering students, it is crucial to explore the variables that contribute to positive academic outcomes in computing ﬁelds, to ﬁnd ways to improve the graduation rate.To develop an understanding of which variables are the most important for students’ undergraduate graduation from computing programs, we apply Astin’s Input-Environment-Output (I-E-O) model to a ﬁltered version of the Multiple-Institution Database for Investigating Engineering Longitudinal Development (MIDFIELD) composed of computing students. We implement Machine Learning (ML) algorithms to predict the reasons behind computing students’ attrition. Then, we leverage the best predictive model to analyze the factors that are most critical for graduation in a computing ﬁeld.