Description and prediction: Knowledge discovery in University databases

Abstract

Data mining methods, including machine learning, have been applied for many years in business contexts and are now receiving a great deal of attention from educators and data scientists in higher education. Accurate, early prediction of whether a student is on track to graduate with distinction is a critical tool for administrators, educators, and advisers to ensure that at risk students are properly supported and students on the path to the highest success are able to stay on track. This study proposes a knowledge discovery in databases approach toward the development and evaluation of several prediction models to accurately predict with two years of academic data or less whether students will graduate with distinction. In developing the prediction models, several important factors of students' success are identified as well as additional insights about student experience.