Idiographic artificial intelligence to explain students' self-regulation: Toward precision education

Existing predictive learning analytics models have exclusively relied on aggregate data which not only have obfuscated individual differences but also made replicability and generalizability difficult. This study takes a radical departure and uses a person-specific approach to predicting and explaining students' self-regulation (SRL). A person-specific approach entails developing a predictive model for each individual using their own data (i.e., idiographic, single-subject or N = 1). We also use explainable and interpretable artificial intelligence (AI) models that allow us to identify the variables that explain students' SRL and guide data-informed decisions. Our study has shown that idiographic single-subject models are tenable, informative, and can accurately capture the individualized students' SRL process. Predictions varied vastly across students regarding accuracy and predictors. The traditional average model did not match any student regarding the predictors' order. These findings are a testament that the "average" is rare and often does not represent any individual student. The variability in our study has shown that no single model can accurately and reliably capture all students. To account for the unique learning processes of individual students, idiographic methods could provide a solution. Educational relevance statement: Individualized artificial intelligence is feasible and reliable and can help understand each person using their own data. Using idiographic models, we can deliver solutions that are precise, accurate and interventions that are more likely to work.