bullet">students’ grades

course selections

attendance records

demographic information

Researchers applied data-mining and visualization techniques to track students’ learning pathways and academic progress over time. The goal was to help institutions understand patterns in student success and improve educational planning.

What this research shows

Educational data mining can be used as evidence to:

identify patterns of student performance
detect risk of failure or dropout
understand how students move through programs
support institutional decision-making

2. What Educational Data Mining Can Tell Us

Educational Data Mining (EDM) analyzes data produced in learning environments to discover meaningful patterns about learning and teaching.

a. Student Learning Behaviors

EDM can reveal:

how often students engage with learning materials
study habits and participation patterns
collaboration and online interaction behaviors

Researchers use clickstream data, assessment results, and platform activity logs to understand learning processes.

b. Prediction of Academic Performance

One major strength of EDM is prediction.

Studies show EDM methods can:

predict exam success or failure
identify students at risk early
estimate future academic outcomes

These predictions allow early intervention and personalized support.

c. Personalised Learning and Teaching Improvement

EDM supports:

adaptive learning systems
curriculum improvement
data-driven teaching decisions

By discovering hidden patterns, educators can tailor instruction to different learners.

d. Institutional and Policy Insights

Large-scale EDM research helps institutions:

evaluate course effectiveness
understand progression pathways
improve retention and graduation rates

Reviews of EDM research identify learning, predictive, behavioural, and visualization analytics as major areas of impact.

3. What Educational Data Mining Cannot Tell Us

Despite its power, EDM has important limitations.

a. It Cannot Fully Explain Why Learning Happens

EDM finds patterns, but patterns ≠ causes.

For example:

Data may show low engagement predicts failure,
but it cannot always explain why students disengage (motivation, family issues, stress, etc.).

Human interpretation and educational theory are still required.

b. It Misses Social and Emotional Factors

Many important learning influences are hard to capture in datasets, such as:

motivation
creativity
confidence
classroom relationships

Data alone cannot represent the full human learning experience.

c. Risk of Bias and Misinterpretation

EDM depends on available data:

incomplete datasets lead to misleading conclusions
algorithms may reinforce existing inequalities
predictions can label students unfairly

Researchers therefore stress ethical and careful use of analytics.

d. Prediction Is Not Certainty

EDM models estimate probabilities, not guaranteed outcomes.

A student predicted to struggle may still succeed, meaning decisions should never rely only on algorithms.

4. Key Challenges of Educational Data Mining

Major challenges identified across EDM research include:

Privacy and data protection concerns
Data quality and integration across systems
Need for educators to understand analytics results
Balancing automation with human judgment
Ethical use of student data

EDM and learning analytics are now evolving toward a broader field called educational data science to address these challenges.

5. Overall Evaluation

Educational data mining provides powerful evidence about learning patterns and educational effectiveness, but it is not a complete picture of education.

In simple terms:

✅ EDM tells us what students are doing and what may happen next.

❌ It cannot fully tell us why students learn, how they feel, or what teaching means socially and emotionally.

The best educational research combines:

data mining,
educational theory,
and human professional judgment.

Educational Data Mining (EDM): Research Evidence, Possibilities, and Challenges