Publication Date

5-2020

Date of Final Oral Examination (Defense)

3-12-2020

Type of Culminating Activity

Dissertation

Degree Title

Doctor of Education in Educational Technology

Department

Educational Technology

Supervisory Committee Chair

Jesús Trespalacios, Ph.D.

Supervisory Committee Member

Jui-Long Hung, Ed.D.

Supervisory Committee Member

Lida Uribe-Flóres, Ph.D.

Abstract

Higher education is attracting more students from diverse background especially at public community colleges. These institutions can help these students attain a quality education at a reasonable price. Unfortunately, community colleges have lower graduation rates than 4-year institutions in part due to the diverse needs and variety in academic preparedness amongst their populations. It can be difficult to identify students most at risk of performing poorly until it is too late. There are multiple ways to predict students’ performance. In this study, three common data mining techniques are compared for their accuracy in predicting academic success using only data collected at the point of admissions. Accurate early prediction can allow academic support professionals to intervene and provide intrusive assistance. A neural network model was found to be more accurate than logistic regression and decision tree models. Moreover, data elements of high school GPA, age, and sex were the most important factors in the neural network model.

DOI

10.18122/td/1653/boisestate

Download

Share

COinS