Predictive Modeling of Student Enrollment in Nepalese Universities: A Case Study Using Statistical and Mathematical Methods

Abstract

Student enrollment is one of the most visible indicators of the expansion and changing demand for higher education. However, counts reported for a single year are not sufficient for planning classrooms, faculty, scholarships, digital services, examinations, hostels, and public financing. This study develops a transparent predictive framework for university enrollment in Nepal by combining descriptive statistics, concentration measures, trend analysis, linear regression, exponential regression, growth rates, and scenario projection. The empirical case study uses official University Grants Commission (UGC) data, with the 2023/24 system-level total of 633,053 students and university-level enrollment observations for selected institutions. The seven focal universities are Tribhuvan University (TU), Purbanchal University (PU), Pokhara University (PoKU), Kathmandu University (KU), Far Western University (FWU), Mid-West University (MU), and Rajarshi Janak University (RJU).

The descriptive results show an exceptionally concentrated enrollment structure: TU accounts for 491,299 students, or 77.61% of the national total reported in the UGC dataset, while the seven selected universities together account for approximately 97.68%. Female students constitute 56.57% of the national enrollment in the reported data, although university-level gender patterns differ. For predictive modeling, the paper uses a longer annual series available for TU and KU from 2015/16 through 2023/24, with the 2022/23 observation unavailable in the selected UGC table. Linear and exponential models are estimated and compared using R² and RMSE, while model parsimony and policy plausibility are considered before selecting a baseline forecasting approach. The resulting forecasts are not treated as certainties; they are planning estimates conditional on historical patterns continuing.

The study finds that simple linear trend models provide a useful conservative baseline, whereas exponential models generate higher forecasts when sustained proportional growth is assumed. For TU, the forecasts diverge substantially, demonstrating why scenario ranges are more informative than a single point estimate. For KU, the models produce much closer forecasts, reflecting a more stable historical trajectory. At the national level, a two-observation growth scenario based on 2021/22 and 2023/24 implies an average annual growth rate of about 4.52%; because only two national observations are used, this is explicitly treated as a scenario rather than a statistically validated forecast. The paper recommends institutional-level longitudinal data, cohort tracking, program-level predictors, and rolling forecast evaluation as the next stage of Nepalese higher-education planning.