Using machine learning to aid in recommending population-level interventions in schools

Disabled students are at particular risk of not attaining strong results in core educational subjects in Key Stage 4. (National Statistics, 2022) This is explainable in part by some disabilities-such as learning disabilities-directly impacting students' ability to learn and engage with the curriculum. However, the impact of marginalisation and other social factors on attainment of particular subjects is well-documented. (Strand, 2014; National Statistics, 2022) Disability is not only a medical phenomenon; it is also created through the construction of 'disabling environments' which disempower individuals and prevent them from acting with agency. The factors that make an environment disabling can be physical and social; often, physical barriers lead to social exclusion. , this research would aim to uncover relationships between subject-specific uptake and attainment and a number of variables known to impact uptake and attainment using machine learning. Machine learning is an appropriate approach for this research because it allows attainment projections to simultaneously be created for every academic subject for a given intersection 'profile'. It is also an excellent way to learn bias in existing data. (Mehrabi, Morstatter, Saxena, Lerman, & Galstyan, 2021) The results for different profiles may then be taken as an indicator of areas where systematic bias affects students, creating a quantitative argument for large-scale change and a metric that can be used to gain insight into changes in the educational attainment of specific subpopulations over time.