Regularly audit ML models for fairness and proactively apply bias mitigation during model training¶

Strategy · All strategies
Evidence · no claims cited

Description¶

This strategy directs practitioners of predictive analytics in education to treat fairness as an ongoing operational concern rather than a one-time check. The article's Notes for Practice state that "Regular audits of machine learning models are crucial to identify biases and ensure equitable outcomes across diverse student populations" and that practitioners "should proactively apply bias mitigation techniques (e.g., reweighting, suppression, disparate impact remover) during model training." It also calls for transparency in preprocessing decisions such as outcome aggregation, since these influence both bias and accuracy.

Design Implications¶

Context¶

Requirements¶

• Access to model performance broken down by key demographic subgroups (e.g., gender, age, disability, socioeconomic status)
• Availability of bias mitigation techniques applicable during model training

Constraints¶

• The article did not evaluate the long-term effects of these mitigation strategies in real-world educational settings

Target Learners¶

• adult learners in virtual learning environments
• higher education students in online courses

Target Learning Goals¶

• equitable identification of at-risk students for timely intervention

Related Strategies¶

• Use this Toolkit: Mitigating Racial Bias in Edtech
• Mitigating Racial Bias in Edtech Products
• AI in Education Toolkit for Racial Equity
• Evaluate AI systems in DLPs for fairness across learner groups, not just overall accuracy
• Deploy ML-based early-warning prediction to shift retention support from reactive to proactive intervention
• Jointly investigate fairness in machine learning with core ML researchers when deploying adaptive collaborative learning support

Examples¶

-

Key Sources¶

• Raymond A. Opoku, Bo Pei and Wanli Xing. (2025). Unveiling Accuracy-Fairness Trade-Offs: Investigating Machine Learning Models in Student Performance Prediction. Journal of Learning Analytics 12(2). https://doi.org/10.18608/jla.2025.8543