Forecasting Enrolment Trends using ARIMA Model: A Multi-Metric Evaluation Approach

Accurate enrollment trend forecasts are essential in education as they facilitate efficient resource allocation, strategic marketing, and optimal budget utilization. These forecasts empower educational institutions to navigate the ever-changing educational landscape with astuteness and foresight. In this research, the ARIMA (Autoregressive Integrated Moving Average) model, one of the most widely used machine learning approaches, is employed to forecast enrollment trends in a state university in the Philippines. The study assesses accuracy and reliability using evaluation metrics such as Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Mean Absolute Percentage Error (MAPE), Prediction of Change in Direction (POCID), Coefficient of Determination (R2), Akaike Information Criterion (AIC), and Bayesian Information Criterion (BIC). Historical enrollment data from the university spanning 30 years was utilized, revealing non-stationarity in the data and highlighting the need for time series analysis. The ARIMA (1,2,0) model performs best, with lower RMSE, MAE, and MAPE. The ARIMA (1,1,0) model provides a concise explanation with the lowest AIC and BIC. The ARIMA (2,0,1) model excels in predicting directional changes (POCID). The optimal choice is the ARIMA (1,2,0) model, considering higher R2 and lower MAPE and RMSE. Forecasted enrollment values demonstrate consistent growth, with an initial surge of over 11 percent followed by gradual moderation to around six percent in the second year and subsequent small increments. Implementing the recommended model improves resource allocation and infrastructure planning, while regular monitoring and refinement enhance forecast accuracy