Transfer Learning-Based Case Base Preparation for a Case-Based Reasoning-Based Decision Making Support Model in the Educational Domain

Abstract

Decision making support in the educational domain is very important for the success of the students, especially for that of the in-trouble students who are asked to stop their study. As a further work of early prediction of the in-trouble students, our current work is thus dedicated to a decision making support model in the educational domain for the problem of study extension of those in-trouble students. Different from the existing educational decision support systems and their models, our model is developed with a combination of case-based reasoning and transfer learning. This combination stems from a more practical context where there are little target data and corresponding target cases available for decision making support. Therefore, our model utilizes case-based reasoning for its problem solving process while making use of transfer learning for case base preparation with not only the limited number of target data but also the larger number of source data. In addition, with the instance-based transfer learning-based method, the case base of our model can be constructed and maintained over the time so that new target cases can be supported with enough similar cases for forming their proper solutions. An empirical study on real data sets has shown that our initial work is promising to have a rich case base for the proposed educational decision making support model.