Published Paper Details:

Forecasting plays a vital role in helping organizations anticipate future demand and optimize decision-making processes. Classical statistical forecasting techniques continue to form the foundation of quantitative analysis within management education, particularly for students beginning their study of analytical methods. This paper presents a comparative evaluation of three fundamental forecasting models: Autoregressive Integrated Moving Average (ARIMA), Exponential Smoothing (ES), and Multiple Linear Regression (MLR). A simulated monthly demand dataset is used to illustrate the application of each model under controlled conditions. The study includes a review of key forecasting literature, theoretical underpinnings of the models, detailed methodology, and accuracy evaluation using Mean Absolute Percentage Error (MAPE), Root Mean Square Error (RMSE), and Mean Absolute Error (MAE). Results show that Holt's Exponential Smoothing performs best for datasets with stable linear trends, followed by ARIMA, while regression captures the deterministic trend but is less responsive to random variations. Simple Exponential Smoothing performs weakest due to structural limitations. The study reinforces the importance of selecting forecasting models that match data characteristics and highlights the continued relevance of classical statistical techniques for introductory management and analytics education.