Data mining and machine learning in education with focus in undergraduate cs student success

Abstract

Computer science (CS) enrollments are at an all-time high, [1] and successful undergraduate CS graduations are indisputably important. With a student population of approximately 51,000, Georgia State University is a USA based state university which is diverse and forms a rich big data footprint as students navigate pathways to graduation. Quoted in a July 2017 article from HigherEd.com, "Georgia State's extensive predictive analytics efforts are leading to better grades and student retention - and more minorities graduating from STEM programs." This doctoral project builds upon current data mining and modeling, machine learning applications, and learning analytics for predicting student success that is beyond retention. Gaining knowledge of CS student learning, developing better alerting models for success, and discovering behavioral indicators from learning analytics reporting is the goal of this research. Using this knowledge as evidence based data for improving the CS student experience will aid in performance improvements and increase pathways to graduation. My supporting research project is building CS student datasets to represent the student as directed graphical models, investigating their relationships using machine learning frameworks, and complex mathematical computations (tensors or gradient boosting) along with graph data mining techniques.