Enrolment Forecasting using Holt’s Double Exponential Smoothing Method with Particle Swarm Optimization

Abstract

Higher education institutions usually do enrolment forecasting to predict the number of students in the future academic year. It is an essential process that plays a crucial role in the effective planning, resource allocation and decision-making activities within the university. In this study, the enrolment was forecasted using Holt’s double exponential smoothing method with the application of particle swarm optimization.  The enrolment data was collected from the historical data of the university registrar. The dataset involved ten years of enrolment data, where six years were allotted for the training set and the remaining four years were used for the test set. Pre-testing of the data showed the linear trend and applicability of the model. The training set was used to compute for the smoothing constants of the Holt’s Double Exponential Smoothing Method where the process was optimized using particle swarm optimization (PSO). The test set was then used to measure the performance of the model. Generally, the model performed well on making forecasts and was able to correctly detect rise and fall of the enrolment data.