Linked data, data mining and external open data for better prediction of at-risk students

Research in student retention is traditionally survey-based, where researchers use questionnaires to collect student data to analyse and to develop student predictive model. The major issues with survey-based study are the potentially low response rates, time consuming and costly. Nevertheless, a large number of datasets that could inform the questions that students are explicitly asked in surveys is commonly available in the external open datasets. This paper describes a new student predictive model that uses commonly available external open data instead of traditional questionnaires/surveys to spot `at-risk' students. Considering the promising behavior of neural networks led us to develop student predictive models to predict `at-risk' students. The results of empirical study for undergraduate students in their first year of study shows that this model can perform as well as or even out-perform traditional survey-based ones. The prediction performance of this study was also compared with that of logistic regression approach. The results shows that neural network slightly improved the overall model accuracy however, according to the model sensitivity, it is suggested that logistic regression performs better for identifying `at-risk' students in their programme of study.