Bayes factors on Gaussian-process trajectories show each behavioural cluster is statistically distinct from the others¶

Claim · All claims
Evidence · 1 study · associational r2 · q2
Currency · newest study 2019 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? GPR Bayes factors confirm that cluster trajectories derive from separate processes (overall K = 3.37×10^10), with pairwise factors between neighbouring clusters of 0.38, 1.52 and 0.17×10^10. → Peach 2019

Evidence¶

Peach 2019¶

Peach, R. L., Yaliraki, S. N., Lefevre, D. and Barahona, M. (2019). Data-driven unsupervised clustering of online learner behaviour. npj Science of Learning 4:14. https://doi.org/10.1038/s41539-019-0054-0

q2 · i? · associational · r2

Gaussian Process Regression analysis of the 6-cluster partition of the 81-learner cohort. Comparing separate cluster log-likelihoods against the whole set, "The Bayes Factor ... was found to be large ( K = 3.37 × 1010)", with pairwise factors between neighbouring clusters also reported.

"The Bayes Factor, calculated as the sum of log likelihoods of each separate cluster minus the log-likelihood of the entire set of trajectories 12 was found to be large ( K = 3.37 × 1010)."

Discussion¶

Related Claims¶

• In pairwise WD analyses, learner trajectory similarity dominates variation in percentage-overlap values (57.90% of variation) while remaining design factors are essentially zero — related
• Type 2 groups showed a communication–behaviour-synergistic transition, Type 1 a behaviour-oriented, and Type 3 a communication-oriented transition in hidden Markov models — related
• In simulated epistemic games, the weighted density statistic distinguishes simulated learner types with distinct mastery trajectories, with the expert trajectory showing the largest WD values — related