Classifying Courses at Scale: a Text as Data Approach to Understanding Student Course-Taking

Abstract

This technical report summarizes the training steps for Classifying Courses at Scale, a project that uses machine learning and natural language processing to classify courses from student transcripts into a standardized hierarchical taxonomy of course-taking called the College Course Map (CCM). Applying our CCM classification models to unstandardized administrative transcript data allows researchers to compare student course-taking patterns across multiple institutions and postsecondary systems. This technical report summarizes the model training steps. Future versions of the technical report will include model evaluation, error analysis, and demonstrations of applications of the models to administrative data.