Bayesian network model of factors related to academic failure in Tehran university students

Researchers have always tried to prevent the formation and escalation of academic failure by making decisions for educational system administrators. For this reason, they have been using forecasting statistical methods. Bayesian network is one of the Predictor Data mining methods and Academic failure is one of the major problems in the higher education system.

The purpose of this study was to identify factors associated with academic failure using educational data mining and Bayesian networks.

The research population was all undergraduate students of Tehran University. A sample of 800 people was selected by Stratified Random Sampling. After performing and eliminating confused questionnaires, 746 questionnaires were analyzed and data were collected using   researcher-made students' academic failure questionnaire, which had a validity of professors and Cronbach's alpha coefficient of 0/971.

Data collected from students was analyzed and Bayesian model was obtained.The accuracy of the algorithm was equal to 95/84% .

The results showed that all factors of the questionnaire were effective in the occurrence of academic failure and individual factors were the only factors that directly and indirectly affect the occurrence of academic failure. The accuracy of the algorithm indicates that the Bayesian network model can predict the academic failure well.