FORECASTING THE NUMBER OF COLLEGE STUDENTS IN THE US USING LONG SHORT-TERM MEMORY

Abstract

The long-term viability of educational institutions typically relies heavily on strategic planning, including effectively distributing resources and personnel, as well as setting aside funds for financial assistance and grants for new students. Therefore, accurately predicting student enrollment is essential for making important decisions based on past temporal data. In this paper, a time series algorithm named \long short-term memory" (LSTM) is utilized to forecast the number of college students in the US. For this purpose, a dataset containing the number of college students in the United States throughout the years is utilized for the training process in the LSTM to acquire an optimal LSTM model. Subsequently, this optimal model is employed to rapidly and reliably forecast the number of students in the upcoming years. The effective- ness and accuracy of the current method are validated by contrasting the obtained results with precise data.