Integrating learning analytics to predict student performance behavior

Abstract

Using Learning Analytics (LA) in educational institutions is an area that has experienced unprecedented growth over the years. LA is the collection and analysis of electronic data to observe hidden patterns in the learning process. One of the main aims of LA is to help faculty and advisors determine which students might be at risk and who are facing difficulty in their academic career. Drawing upon extant literature, this paper proposes and discusses the development of a new prediction model. The proposed model takes the advantage of the fully electronic characteristics of student data, which include student activity and their marks. Compact Prediction Tree (CPT+) has been described in literature and has proved its efficacy to solve various problems in many different disciplines. In this paper, a new prediction model based on a CPT+ algorithm to predict the next grade for upcoming courses or for the registered course(s), is proposed. To evaluate the performance of the proposed model, dataset in CAS were applied as the test problem set. The method was examined in terms of accuracy and its result was significantly better than other prediction models, namely Dependency Graph (DG) and Prediction by Partial Matching (PMP).