engleski

Predictive Modelling of Academic Performance by Means of Bayesian Networks

Predicting academic performance is an often- required task in Higher Education field. Development of data mining, especially educational data mining (EDM) provided algorithms for effective data analysis with the aim to improve quality of the educational processes. In this paper, probability based approach to machine learning (Bayesian networks) is applied in order to predict academic performance of IT students based on data about their socio- demographic characteristics, attitudes, motivation and behavior. Main aim of presented research was twofold: (i) to predict students’ academic performance and to identify most significant predictors of students` success, (ii) to investigate possibilities of probability based machine learning approach for developing predictive models in educational domain. Research results indicated high level of potential for Bayesian networks application on educational datasets.

bayesian networks, academic success, data mining, CRISP DM process model

nije evidentirano