Using non-identifiable data to predict student course selections

Publication date: Available online 10 December 2015 Source:The Internet and Higher Education Author(s): Ivana Ognjanovic, Dragan Gasevic, Shane Dawson The ability to predict what university course a student may select has important quality assurance and economic imperatives. The capacity to determine future course load and student interests provides for increased accuracy in the allocation of resources including curriculum and learning support and career counselling services. Prior research in data mining has identified several models that can be applied to predict course selection based on the data residing in institutional information systems. However, these models only aim to predict the total number of students that may potentially enrol in a course. This prior work has not examined the prediction of the course enrolments with respect to the specific academic term and year in which the students will take those courses in the future. Moreover, these prior models operate under the assumption that all data stored within institutional information systems can be directly associated with an individual student's identity. This association with student identity is not always feasible due to government regulations (e.g., student evaluations of teaching and courses). In this paper, we propose an approach for extracting student preferences from sources available in institutional student information systems. The extracted preferences are analyzed using the Analytical Hierar...
Source: The Internet and Higher Education - Category: Information Technology Source Type: research