First International Online Seminar on Computational Educational Sciences

Center for Computational Educational Sciences of HSE University eLearning Office and personally Dr. Dmitry Abbakumov are holding the 1st International Online Seminar on Computational Educational Sciences.

Carlos Felipe Rodriguez Hernandez (KU Leuven, Belgium) will present his research "An artificial neural network analysis of academic performance in higher education” (full text).

Abstract:

Although the use of predictive systems based on artificial neural networks (ANNs) is well known in several research areas, it has received less attention within the educational research field. As a matter of fact, both conceptual and methodological limitations have prevented the use of ANNs for the prediction of students’ academic performance. To address such limitations, the present study proposes a multistage approach for the use of ANNs to predict the academic performance of 162,030 university students. Three stages are followed, namely, design, evaluation, and interpretation. Findings suggest that it is possible to systematically train and test neural network models to classify students’ academic performance as either high-performance (accuracy of 82%) or low-performance (accuracy of 75%); to assess the quality of the derived predictions using several quality measures; and to provide clear and transparent interpretations of the predictive models based on ANNs. Furthermore, prior academic achievement, socio-economic conditions, and high-school characteristics continue to be weighty predictors of students’ academic performance in higher education.

The seminar will take place on April 9, Thursday at 18:00 online using a webinar service. 

Working language is English.