In exploring how educational data mining contributes to evidence-based teaching, I reviewed a study by Baker & Yacef (2009), a foundational work in the EDM field. Their research shows how EDM can analyze students’ behaviors within digital learning environments to identify patterns such as off-task activities, guessing, or productive learning strategies. For example, using log data and machine learning models, the study was able to detect when students were struggling or disengaged, allowing instructors to intervene more effectively.

What EDM can tell us:

Patterns of engagement and disengagement

Predictive indicators of student success or risk

Learning behaviors such as persistence, hesitation, or guessing

How students navigate digital content and assessments

What EDM cannot fully explain:

The emotional reasons behind student choices (e.g., stress, motivation, external issues)

The quality of learning or depth of understanding without additional qualitative evidence

Contextual factors like teaching style, home environment, or personal challenges

The nuance of student creativity, collaboration quality, or reflective thinking

Although EDM provides powerful insights, it becomes truly meaningful only when combined with teacher judgment and human-centered interpretation.

Reference:
Baker, R. S., & Yacef, K. (2009). The State of Educational Data Mining in 2009: A Review and Future Visions. Journal of Educational Data Mining.
(Available here: https://jedm.educationaldatamining.org/index.php/JEDM/article/view/8)