A Performance-Based Comparison of Time Series Models for Champagne Sales Prediction

Abstract

Time Series Analysis (TSA) is a critical analytical tool for analyzing data points gathered or captured at predetermined time intervals. It has proven vital in a variety of sectors, particularly financial services, environmental studies, and medicine. It emphasizes the need to handle missing values, smoothing data, and converting non-stationary data to stationary formats. The conducted research dives into the complicated procedures and strategies used in TSA, covering an extensive variety of models including Autoregressive Integrated Moving Average (ARIMA), Autoregressive Integrated Moving Average with Exogenous Variables (ARIMAX), and Seasonal Autoregressive Integrated Moving Average (SARIMA). Enterprises may improve decision-making, streamline procedures, and predict potential patterns by employing strong TSA. The research paper worked on a case study highlighting SARIMA’s superior performance in circumstances where seasonality is a substantial factor, making it a preferred choice over ARIMA for TS forecasting involving seasonal data. The study investigates and compares the efficiency of the ARIMA and SARIMA approaches for TSA to evaluate the model's effectiveness using performance metrics like Mean Absolute Error (MAE), Mean Square Error (MSE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE). SARIMA outperforms ARIMA in standings of all performance metrics because of its ability to explicitly model seasonality in TS data.