2026 ASEE Annual Conference & Exposition

Unequal distribution of educational opportunities and resources continues to lead to unequal opportunities for K-12 students to participate in computing and artificial intelligence (AI) learning. To address this challenge, our study proposes an interpretable machine learning (ML) framework to investigate the impact of structural factors and contextual factors on student learning outcomes across different schools and districts.
This framework integrates data from the Civil Rights Data Collection (CRDC), which captures systemic inequalities in funding, staffing, and technology access, and the data from EdNet, which is a large-scale behavioral dataset that reflects student engagement and persistence in digital learning environments.
Differentiating from traditional black-box models, our study employs a random forest algorithm and Local Interpretable Model Agnostic Explanations (LIME) to highlight the role of funding levels, student-teacher ratios, technology accessibility, and demographic factors such as race, gender, and socioeconomic background in influencing learning outcomes.
Our ongoing analysis aims to identify how funding gaps, student-teacher ratios, and technology access contribute to patterns of distribution of educational resources. By linking macro-structural indicators with micro-level learning behaviors, the framework aims to provide transparent, evidence-based insights for policymakers, researchers, and education leaders.
In the future, we plan to expand our macro-structural indicators and micro-level learning behaviors to have a thorough study. to promote the responsible application of artificial intelligence in the education system, reveal the drivers of systemic inequality that traditional black box models often ignore, and provide data support and methodological basis for achieving educational equity and optimal resource allocation.