Predicting Who Will Graduate

ISBN-13: 9783639140231 / Angielski / Miękka / 2009 / 152 str.

John Campbell

Opis

Predicting who will graduate from a university is adifficult challenge, especially for US publicuniversities whose missions serve diverse populationsunder relaxed admission criteria. Building predictivemodels for entering freshmen poses many problems:some students receive financial aid, others do not;some enter with SAT scores, others with ACT scores;some students stop out and then return. And, with theadvent of the modern data warehouse, a dizzying arrayof data exists, which might, or might not, help buildpredictive models. This doctoral study examines thework required to build four predictive models forentering freshmen: logistic regression, automaticcluster detection, neural network, and decision tree.Practical problems are addressed squarely: Cleaninginstitutional data, dealing with missing data,adjusting model parameters, recognizing model drift,grouping students into prediction bands, andevaluating disparate model types are just some of thepractical solutions shared in this work.

O autorze

John is Director of Academic Computing at Northern Arizona
University. He entered college, unknowingly, with risk factors
against graduating; but graduate he did, in math, then
engineering, and finally with an EdD in educational leadership.
Here he combines his interests in education and open source
software to build graduation prediction models.