A three-stage filtering pipeline (internal, association, and cluster-quality metrics) reduced 20,724 tested clustering models to select the optimal model per analysis¶

Claim · All claims
Evidence · 1 study · associational r2 · q2
Currency · newest study 2025 · not yet checked against its sources · not reviewed by a person

Subclaims¶

q2 i? 20,724 clustering models were tested, 3,454 per analysis, and filtered through internal metrics, association metrics, and cluster quality metrics. → Fabian Santiago-Muñoz 2025

Evidence¶

Fabian Santiago-Muñoz 2025¶

Fabian Santiago-Muñoz, Mikel Larrañaga, Jon A. Elorriaga, Ana Arruarte. (2025). Identification of Patterns to Prevent the Failure of Engineering Students. Int J Artif Intell Educ 35:3356–3405. https://doi.org/10.1007/s40593-025-00502-9

q2 · i? · associational · r2

Experiment Results section reports the exhaustive hyper-parameter exploration: "20,724 clustering models" were tested, 3,454 per each of the six analyses, before internal-metrics filtering removed 5,988 models.

"As can be seen in Table 8, 20,724 clustering models (i.e., clustering algorithm with a particular setting) were tested in order to get the optimal model for each analysis, 3,454 models for each analysis (RQ1_IA, RQ1_IC, RQ2_IA, RQ2_IC, RQ3_IA, RQ3_IC)."

Discussion¶

Related Claims¶

• The Gaussian Mixture algorithm achieved optimal clustering solutions in four of six scenarios, while no evaluated algorithm yielded satisfactory results in the remaining two — related
• The multi-stage filtering pipeline reduces Beyond-K-5 retention to near zero while preserving a minority of K-5 content — related