Understanding the learning of disabled students: An exploration of machine learning approaches

Research is limited in demonstrating how learning analytics can support students with a learning disability. It is vital to scrutinise the demographics of students with and without disabilities and their effect on performance. Machine learning algorithms can provide valuable insights by mining this expanding education data. This study aims to analyse the relationship between disabled and non-disabled students' demographic with their scores and the number of attempts to complete a module to develop prediction models. Three models - Adaptive Boosting, Random Forest, and K nearest neighbour - were used. The results found that the Adaptive boosting algorithm delivered the highest prediction accuracy.