Predicting the Enrollments in Humanities and STEM Programs in Higher Education Using ARIMAX Models

Abstract

Traditionally, the participation patterns in the humanities and STEM (science, technology, engineering, and mathematics) programs in higher education differ. This study aimed to tackle this issue using concurrent time series data sets in the expanding higher education system. Authors selected the higher education system in Taiwan as an example. The participation in the humanities and STEM programs, covering 71 periods from 1950-2020, were collected from the Ministry of Education in Taiwan. The authors applied CCF (cross-correlation function) and ARIMAX (multivariable autoregressive integrated moving average) models to select the fittest model to predict the future trend. The humanities was the input variable and STEM was the output variable in the model. The findings revealed that ARIMAX (1,2,1) works well for these target data sets. According to the findings, enrollment in STEM programs will decrease with the decline in humanities programs in the future. This finding may provide useful information for related policy makers.