Forecasting Seasonal and Trend-Driven Data: A Comparative Analysis of Classical Techniques

Making future predictions based on past and present data is known as forecasting. In the face of uncertainty, organizations rely on this valuable tool to make informed decisions, develop better strategies, and become more proactive. This study presents a comprehensive comparison of the performance of several classical quantitative forecasting methods, namely, Moving Average, Single Exponential Smoothing, Holt’s Double Exponential Smoothing with a trend, Holt-Winter’s Triple Exponential Smoothing with a trend and seasonality, ARIMA, ARIMAX, SARIMA, SARIMAX, and Multiple Linear Regression method.This research’s aim is to identify the most effective technique for predicting weekly sales of a product, a critical aspect of supply chain management, with the emphasis being placed on the capability of each technique to capture the trend and seasonality components of the dataset. For this, an out-of-sample validation procedure was used; the evaluation of the performance of each technique’s model was conducted using three accuracy metrics: Mean Absolute Percentage Error (MAPE), Mean Absolute Error (MAE), and Root Mean Square Error (RMSE). The results revealed that the SARIMAX model outperformed the other techniques, providing the most accurate forecasts for the product’s weekly sales. This paper contributes to the field of industrial engineering by offering insights into the application of these classical quantitative forecasting methods in real-world scenarios, particularly in sales forecasting. The findings of this study can assist businesses and organizations in making up-to-date decisions and developing more effective and successful strategies.