Prediction algorithms forecast student achievement from small windows of data, more accurately when data are enriched with instructional design and learning-process detail¶

Claim · All claims
Evidence · 1 study · design r1 · q1
Currency · newest study 2026 · not yet checked against its sources · not reviewed by a person · recheck by 2029: tool and model accuracy changes quickly

Subclaims¶

q1 i? The article asserts, citing the learning analytics literature, that prediction algorithms can predict student achievement with small windows of data intake, that they are more accurate when raw data are enriched to reflect instructional design and learning processes, and that predictions can be made within days when context-responsive temporal coding is applied. → prediction-accuracy

Evidence¶

prediction-accuracy¶

Bernacki, M.L., and Plumley, R.D. (2026, February). CLICKSTREAM: Essential educational infrastructure to promote student learning and educational theory development. Digital Promise. https://doi.org/10.51388/20.500.12265/285

q1 · i? · design · r1

The white paper attributes these assertions to prior studies (Bernacki et al., 2020a; Plumley et al., 2024) rather than reporting new analyses; no effect sizes or test statistics are printed here. The claim is recorded as the article states it, citing NAME.

"These prediction algorithms can be more accurate when raw data are enriched to reflect instructional design and learning processes (Bernacki et al., 2020a). Predictions can be made within days when context-responsive temporal coding is applied (Plumley et al., 2024)."

Discussion¶

Related Claims¶

• Instructor co-designed tagging transforms raw LMS log data into structured event data reflecting course design and temporal context — a narrower finding that bears on this claim
• Students raised concerns that performance forecasts cannot capture personal circumstances affecting study — related
• Both prior performance flags and the machine learning algorithm are less accurate when predicting outcomes for students who are Black — related
• Enriched DLP data enable two types of theory testing: sustained observation of individuals in a single setting and modeling the same type of learner across contexts — related
• Sequence analysis of online course logs found group coordination predicted successful outcomes better than raw activity levels (review-attributed to Hoppe et al., 2021) — a narrower finding that bears on this claim
• Stealth assessment in AI-GBL achieves accurate prediction of learner states, with bias auditing demonstrated without accuracy loss — a narrower finding that bears on this claim
• Predicting student outcomes was a common AIEd use in higher education, with ten studies predicting academic performance — related
• Course-level sample truncation can produce M-bias that helps explain difficult-to-interpret negative associations between LMS behaviour and student success — related
• The SREM predicts time to reclassification more accurately than a conventional discrete-time hazard model — related
• Seeing an algorithm perform makes people significantly less likely to choose it over a human forecaster — related
• Evidence is mixed on whether including demographic variables as predictors improves model performance in educational prediction — related