Replication data and code: Symmetry-Aware Learning Analytics for University Dropout Risk — Invariant Academic Capital and Rotating Retention Drivers across Four Chilean Cohorts

Program-cohort-level data (100 units = 25 programs x 4 cohorts, 2021-2024; thirteen standardized dropout-risk indicators) and Python code reproducing the cross-cohort configural-invariance analysis (RV coefficient, mean RV ~ 0.65) and the per-cohort HJ-Biplot planes for the article submitted to the Symmetry (MDPI) Special Issue. No individual student records are included; the underlying student-level microdata are confidential.