Forecasting Undergraduate Majors: A Natural Language Approach (AERA OPEN)

Summary:  View help for Summary This Repository contains the code related to the AERA Open  Major Forecasting Paper 
Abstract:
This Repository contains the code related to the AERA Open  Major Forecasting Paper

Abstract: Commitment to a major is a fateful step in an undergraduate education, yet the relationship between courses taken early in an academic career and ultimate major issuance remains little studied at scale. Using transcript data capturing the academic careers of 26,892 undergraduates enrolled at a private university between 2000 and 2020, we describe enrollment histories using natural-language methods and vector embeddings to forecast terminal major on the basis of course sequences beginning at college entry. We find that (I) a student's very first enrolled course predicts their major thirty times better than random guessing and more than a third better than majority-class voting, (II) modeling strategies substantially influence forecasting metrics, and (III) course portfolios vary substantially within majors, such that students with the same major exhibit relatively modest overlap.


Due to the PII nature of the data as well as to protect the underlying institution, data will not be available in this repository. 
If interested in obtaining the data, contact Mitchell Stevens  stevens4@stanford.edu.

see the readme in the code folder for additional details on how to execute the code.

Subject Terms:  View help for Subject Terms major forecasting; higher education; natural language processing; word2vec; course2vec