<![CDATA[Demand forecasting is an important factor to support strategic decision-making, especially in industries with high levels of fluctuation such as motorcycle sales in Indonesia. This study aims to conduct an analysis, compare, and validate the performance of time series methods, namely Moving Average (MA), Weighted Moving Average (WMA), and Exponential Smoothing (ES), to forecast motorcycle sales. The data used are monthly sales data for the 2022–2024 period obtained from the Indonesian Motorcycle Industry Association (AISI). Evaluation was carried out using Mean Absolute Deviation (MAD), Mean Squared Error (MSE), and Mean Absolute Percentage Error (MAPE) with a uniform evaluation period to ensure an objective comparison. The results of the study show that the Weighted Moving Average (WMA) method has the best accuracy on historical data with a MAPE value of 14.88%. Next, validation was carried out using actual data from 2025 with an out-of-sample forecasting approach without model updates. The validation results showed that the Exponential Smoothing (ES) method had the best performance with a MAPE value of 8.81%, indicating better model generalization ability for predicting future data. This study showed that the method with the best accuracy on historical data does not always provide the best performance on future data. Therefore, selecting a forecasting method needs to consider a balance between accuracy and model stability. The results of this study are intended to serve as a reference for selecting a forecasting method suitable for fluctuating data conditions.]]>
Copyrights © 2026
