Repairing Model-Drift in Enrollment Management

Autor/inn/en	Rodriguez, AE; Rosen, John
Titel	Repairing Model-Drift in Enrollment Management
Quelle	In: Research in Higher Education Journal, 43 (2023), (13 Seiten) PDF als Volltext kostenfreie Datei Verfügbarkeit
Sprache	englisch
Dokumenttyp	gedruckt; online; Zeitschriftenaufsatz
Schlagwörter	Models; Enrollment Management; School Holding Power; Data; Algorithms; Sampling; Statistical Bias; Research Problems; Statistical Distributions; College Admission + Suchen Sie Ihr Suchwort? Analogiemodell; Daten; Algorithm; Algorithmus; Forschungskritik; Wahrscheinlichkeitsverteilung; Hochschulzugang; Hochschulzulassung; Zulassung
Abstract	The various empirical models built for enrollment management, operations, and program evaluation purposes may have lost their predictive power as a result of the recent collective impact of COVID restrictions, widespread social upheaval, and the shift in educational preferences. This statistical artifact is known as model drifting, data-shift, covariate-shift. Succinctly, these events drove changes in the stationarity of the target variable and the predictors. The result is a student body with unknown performance qualities entirely distinct from previous cohorts. This study explains and illustrates: (1) how to test for academic model drift in academe; and (2) sets forth two methods used to repair vitiated student-body performance properties. Formally, it frames the data-drift outcome as a One Class problem which allows the deployment of two well-known One-Class algorithms: Support Vector Machines and Isolated Random Forests. The study shows their use in reconstructing a representative sample of the student-body (As Provided).
Anmerkungen	Academic and Business Research Institute. 147 Medjool Trail, Ponte Vedra, FL 32081. Tel: 904-435-4330; e-mail: editorial.staff@aabri.com; Web site: http://www.aabri.com
Erfasst von	ERIC (Education Resources Information Center), Washington, DC
Update	2024/1/01