This research paper focuses on using a convolutional neural network to assess student performance and addresses the impact of the COVID-19 pandemic on education. It introduces a two-step system that combines robust Bayesian model averaging with a frequentist approach for estimating parameters in a multinomial logistic regression model. The authors provide an empirical example illustrating the application of this system in analysing student performance. They also explore strategies to improve e-learning tools by addressing technological factors. The paper contributes to educational evaluation and policy analysis by incorporating deep learning systems and addressing the challenges posed by the pandemic.

Keywords: machine learning; student performance; Bayesian inference; e-learning platforms; logistic regression; variable selection procedure.

JEL Classification: I25.

Cite this chapter:

Pacifico, A. and Giraldi, L., and Cedrola, E.  (2023). Student Performance in E-learning Systems: An Empirical Study. In L., Nicola-Gavrilă (Ed), Digital Future in Education: Paradoxes, Hopes and Realities (164-189 pp.). ISBN: 978-606-95516-1-5. Book Series Socio-Economics, Research, Innovation and Technologies (SERITHA) ISSN: 3008-4237. https://doi.org/10.57017/SERITHA.2023.DFE.ch8

Chapter’s history: 

Received 11th of May, 2023; Revised 17th of June, 2023; Accepted for publication 20th of July, 2023; Published 30th of September, 2023.