A Data-Analytic Approach to Early Detection of At-Risk Students in Higher Education

The use of big data has become increasingly important for fostering innovation and big data analytics is becoming an emerging topic in the field of education, helping educators identify at-risk students who are struggling to graduate. This study undertakes the task of predicting the academic performance of students in higher education using lasso regression. By analyzing large data sets containing records collected early in academic programs, we can identify at-risk students and predict grade point average (GPA) results at graduation with a high degree of accuracy. Additionally, our use of lasso regression modelling helps pinpoint impactful predictors of academic performance, including grades in specific courses in the program that most strongly affect GPA. Results from our analysis are compared with those of other types of regression to demonstrate the advantages of our model over others. This research makes contributions to the field of educational research by offering a dependable and adaptable method of predicting academic performance and a resource for the planning of interventions aiming to provide diverse benefits to support at-risk students’ learning.