Agent-Based Simulation, Machine Learning, and Gamification: An Integrated Framework for Addressing Disruptive Behaviour and Enhancing Student and Teacher Performance in Educational Settings

Abstract

The classroom environment is a major contributor to the learning process in schools. Young students are affected by different factors in their academic progress, be it their own characteristics, their teacher’s, or their peers’. Disruptive behaviour, in particular, is one of the main factors that create challenges in the classroom environment, by hindering learning and effective classroom management. To overcome these challenges, it is important to understand what causes disruptive behaviour, and how to predict and prevent it. While Machine Learning (ML) is already used in education to predict disruption-related outcomes, there is less focus on understanding the processes leading to the effect of disruptive behaviour on learning. Thus, in this thesis, I propose using Agent-Based Modelling (ABM) for the simulation of disruptive behaviour in the classroom, to provide teachers with a tool that helps them not only predict, but also understand how classroom interactions lead to disruptions. Reducing negative factors in the learning environment, like disruptive behaviour, is further supported by increasing positive factors, such as motivation and engagement. Therefore, the use of gamification is then introduced as a strategy to promote motivation and improve engagement by making not only the learning environment more rewarding, but also the ABM teacher simulation more appealing.
This thesis focuses on these issues by designing and implementing for the first time an integrated approach that combines ABM and ML with gamification to simulate classroom interactions and predict disruptive behaviour. The ABM models the complex interactions between students, teachers, and peers, providing a means to study the processes leading to behavioural issues. Meanwhile, ML algorithms help predict learning outcomes with behaviours such as inattentiveness, hyperactivity, and impulsiveness.
The simulation has revealed insights, such as the impact of peer influence on student behaviour and the varying effects of different types of disruptive behaviour, such as inattentiveness, hyperactivity and impulsiveness, on academic performance. The improved performance of the hybrid ML-ABM is shown by measuring results of simulation with ML integration using metrics like MAE, RMSE and Pearson correlation. Moreover, the inclusion of gamification elements was shown to improve engagement by increased login frequency and course completion rates in a MOOC setting, as well as be effective and appealing for teachers using the ML-ABM.
In conclusion, this thesis presents the first comprehensive model that integrates ABM, ML, and gamification elements to explore educational outcomes in a disruptive classroom; it develops the first hybrid ML-ABM approach for predicting and managing classroom disruptive behaviour; it provides empirical evidence of the effectiveness of gamification in boosting student and teacher engagement; and it offers practical insights for educators and policymakers seeking to adopt innovative, technology-driven strategies for improving teaching and learning. The research lays a foundation for future studies, aiming to further explore and expand the capabilities of these technologies in an educational context.