Use data mining in education research in a principled, goal-driven way informed by domain knowledge¶

Strategy · All strategies
Evidence · no claims cited

Description¶

The article's implications section recommends that researchers "be principled in their use of this approach": methods and interpretation should be informed by what is known in the field, the specific objective (prediction, theory development, or hypothesis generation) should determine the method, and special attention should go to sampling, over-fit avoidance, and predictor set completeness. It also urges researchers to maintain a balanced perspective and report null results and unintended downstream consequences.

Design Implications¶

Context¶

Requirements¶

• Requires domain knowledge: mining data "with hardly any knowledge of the domain" is likely to hinder the field

Constraints¶

• The verdict is still out on data mining's usefulness in educational research; convincing applications remain rare even in learning analytics and EDM

Target Learners¶

• education researchers using or considering data mining

Target Learning Goals¶

• principled application of data mining to education research questions

Affordances¶

• Data Mining Versus Traditional Statistics Framework

Related Strategies¶

• Diversify AIED: use diverse technologies, address big problems, and reuse existing resources rather than reinventing content

Examples¶

-

Key Sources¶

• Iwatani, Emi. (2018). Overview of Data Mining's Potential Benefits and Limitations in Education Research. Practical Assessment, Research & Evaluation, 23(15). http://pareonline.net/getvn.asp?v=23&n=15