K-means clustering on EFA-derived features yields six learner groups with distinct usage patterns of UDL-guided elements¶
Claim · All claims
Evidence · 1 study · associationalr2·q2
Currency · newest study 2024 · not yet checked against its sources · not reviewed by a person
Subclaims¶
Reading the evidence codes
q is the design tier, i the impact, n the sample, and the kind with its r
the rigour judged within that kind: what each code means.
This claim: no standardized effect is recorded in observations/ for it yet, so no benchmark reading is possible.
q2 i? Using silhouette score, BIC and AIC, the optimal k-means solution had six clusters, with a silhouette score of 0.506 and cluster sizes ranging from 7 to 226 learners. → Marvin Roski 2024
Evidence¶
Marvin Roski 2024¶
Marvin Roski, Ratan Sebastian, Ralph Ewerth, Anett Hoppe, Andreas Nehring. (2024). Learning analytics and the Universal Design for Learning (UDL): A clustering approach. Computers & Education 214 (2024) 105028. https://doi.org/10.1016/j.compedu.2024.105028
q2 · i? · associational · r2
Unsupervised k-means clustering (scikit-learn defaults, ten initializations, maximum 300 iterations) was run on four weighted-sum-score features derived from the EFA factors for 384 learners. The silhouette score for the six clusters was 0.506; the largest cluster contained 226 learners (58.9%) and the smallest 7 (1.8%).
"When comparing the silhouette score and BIC/AIC at different numbers of clusters, the silhouette score shows a maximum at k = 6, and the BIC/AIC shows a minimum at the same number of clusters"
Discussion¶
Related Claims¶
- The six clusters differ substantially in usage duration of UDL-guided elements, with the small Cluster 2 investing particularly more time than Cluster 1 — related
- The largest cluster of learners invests minimally in the platform and shows almost no gain in conceptual knowledge — related
- Exploratory factor analysis of 384 learners' log data identifies four factors in usage of UDL-guided elements, indicating learners remain consistent in decisions such as video vs. text and tasks with or without help — related
- K-Means++ clustering of self-reported behavioral proxies yields six distinct, stable learner profiles (k = 6, Silhouette = 0.62) — related
- K-means clustering of K–8 students' platform trace data yields seven distinct engagement profiles — related
Page metadata
| Field | Value |
|---|---|
type |
claim |
id |
kmeans-six-learner-clusters-udl-elements |
title |
K-means clustering on EFA-derived features yields six learner groups with distinct usage patterns of UDL-guided elements |
description |
K-means clustering on EFA-derived features yields six learner groups with distinct usage patterns of UDL-guided elements |
status |
review |
evidence_strength |
moderate |
newest_evidence |
2024 |
generated |
by process:wiki-ingest on 2026-10-29 |
trust |
unverified — no one has confirmed this page's content |
sources |
1 — marvin-roski-2024 |
Frontmatter is the page's machine-readable record — see the Schema & Guide.