Paper Title
IDENTIFYING HETEROGENEOUS OUTCOMES WITH MIXED-EFFECT RANDOM FORESTS

Abstract
Abstract - The purpose of this study was to apply a multilevel extension of the random forest, the mixed-effects random forest (MERF), to predict the 9th grade on track to graduation (9G-OTG) status of students in the Pacific Northwest region of the United States using data from Oregon’s state longitudinal data system (SLDS). MERFs were used to account for the longitudinal and higher order nesting of the SLDS data structure and to identify patterns among the large number of variables contained in the dataset, including multi-term interaction effects, and nonlinear associations between predictors and the on-track to graduation outcome. In addition, we demonstrate the use of modern data visualization techniques to clearly highlight the complex functional forms and interactions of variables which underlie the differential student outcomes that were observed. Keywords - Educational Outcomes, Machine Learning, Random Forests, Nested Data Structures.