<![CDATA[Inventory management is an important aspect in maintaining the effectiveness of business operations, particularly in building material stores where demand fluctuations can affect stock availability. Inaccurate inventory planning may lead to overstock or stockout conditions, resulting in increased operational costs and reduced customer satisfaction. This study aims to develop a demand forecasting system for building materials using the Long Short-Term Memory (LSTM) method based on historical sales data at Toko Bangunan Beu Sukses. The dataset used consists of daily sales data from January 2024 to October 2025 covering six products, namely steel, cement, paint, pipes, zinc roofing, and plywood. Data preprocessing was performed through logarithmic transformation, differencing, normalization using MinMaxScaler, and sequence formation using the sliding window method. The LSTM model was trained and evaluated using Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), and Mean Absolute Percentage Error (MAPE). The evaluation results indicate that the proposed model achieved a high level of forecasting accuracy, with all products obtaining MAPE values below 2%. Furthermore, the developed model was successfully integrated into a web-based application to support inventory management and decision-making processes. The results demonstrate that the LSTM method can effectively predict building material demand and support more efficient inventory management. ]]>
Copyrights © 2026
