Content area
Higher education institutions are struggling to maintain enrollment amidst changing higher education landscape and demographics. New programs often represent a means to boost enrollment for institutions and therefore maintain financial stability, but pose a significant risk in terms of the associated upfront investment. Current research has established many methods for forecasting overall enrollment figures for strategic planning and have typically relied on historical enrollment figures and occasionally incorporated external data points. However, current research has not addressed forecasting for institutions looking to establish new programs, which have no historical data available. This study used IPEDS data to develop predictive models for predicting degree completions for new undergraduate degree programs to determine whether a program will be successful. The results indicated significant relationships between degree completions and many independent variables. The response variable was zero-inflated, with about half of the new programs having zero completions in the fourth award year. The predictive analysis developed several random forest and linear regression models. The assumptions for linear regression were not met. The random forest model showed poor fit, with an RMSE of 8.03 and a tendency to overestimate. This study exposed the challenges in predicting degree completions using IPEDS data. Institutions can avoid handling zero-inflated data by seeking alternative indicators for program success. Future research should explore how IPEDS data can be incorporated into planning processes and alternative approaches to determine whether a new program will be successful.