Student Performance Score Prediction Using Artificial Neural Network with the Support of Exploratory Factor Analysis and Clustering

This study proposes a sequence of data analysis techniques for prediction purposes which consists of the use of exploratory factor analysis for classification, clustering of variables in a factor for focusing to a prediction goal and prediction of missing data via an artificial neural network using performances of engineering students' data. Average end-year scores of courses have been used to represent performance of a student. Prediction goal is the forecasting of a course score before the examination. Factor analysis has been used to group courses under individual factors. The factor in which course score to be predicted exists is clustered to group students with similar scores from various courses in that factor, excluding the course score to be predicted. An artificial neural network is trained for course scores in the cluster and target course score of a student is simulated in the artificial neural network for prediction after completion of training with an acceptable training error. The study showed a significant accuracy for prediction. In addition, the realized work has shown that it is possible to re-evaluate an education program, guide a student for better academic performance and provide support for employment via a combination of data mining methods. Python has been used to carry out the analysis.