Applying Machine Learning Approaches with Administrative Data to Investigate Student Level Outcomes: An Application with Kindergarten Entry Assessments

• In Event: AERA Roundtable Session Friday 7:45 am - Gold 4
  In Roundtable Session: RT4: Modeling Complex Data and Patterns: Bayesian, Longitudinal, and Machine Learning Approaches

Abstract

The proliferation of educational data at school, district, and state levels, along with improved access to these data sets, provides opportunities for educational leaders to make well-informed decisions. However, the ability to analyze and interpret these data into actionable steps is difficult given the size and scope of the information available. Machine learning techniques are appealing for use in educational data because they can handle large datasets (e.g., online learning or administrative sources) and large numbers of variables (e.g., demographic data) (Hilbert et al. 2021). Preliminary results in this paper show ridge regression to be more effective at predicting student outcomes than ordinary least squares or lasso regression, but future work includes testing more algorithms.

Authors

• Kristen E. Wright, University of North Carolina at Charlotte

• Kyle T. Cox, University of North Carolina - Charlotte

• Richard G. Lambert, University of North Carolina - Charlotte