Preliminary communication
https://doi.org/10.17818/EMIP/2023/2.3
STUDENT SUCCESS PREDICTION USING ARTIFICIAL NEURAL NETWORKS
Teo Ljubičić
orcid.org/0000-0002-1543-7729
; A1 Hrvatska d.o.o.
Marko Hell
orcid.org/0000-0003-1685-3561
; University of Split, The Faculty of Economics, Business and Tourism
*
* Corresponding author.
Abstract
Many years of electronic data processing have enabled the storage of a large amount of data that can be used today to improve educational processes through machine learning algorithms. Using data from the Moodle distance learning system, an artificial neural network model was created to predict the final outcome of students at the end of their studies based on their final grades of the first year of study. In three artificial neural network models, the power of this algorithm was demonstrated, where all models achieved a very low error, and the artificial neural network model achieved the best results with two hidden layers of nine neurons, whose absolute error was 0.1920, and the squared error 0.0562. The research shows that artificial neural networks are very effective in predicting the final outcome of students based on the grade from the first year of study and that such models have the potential to become an auxiliary tool and means of decision-making in educational institutions.
Keywords
machine learning; deep learning; artificial neural networks; predictive modelling; education system
Hrčak ID:
310853
URI
Publication date:
7.12.2023.
Visits: 880 *