<![CDATA[Inventory management remains a critical challenge for small and medium-sized retail businesses, including Gonvapestore, a vape retailer in Bali, where stock decisions are often made intuitively. This study aims to design and implement a stock forecasting system using the Long Short-Term Memory (LSTM) algorithm to enhance the accuracy of monthly inventory predictions. The research follows the Knowledge Discovery in Database (KDD) process, encompassing data selection, preprocessing, and time series transformation through a sliding window approach. The LSTM model was developed using TensorFlow and Keras, and its forecasting accuracy was evaluated using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), and Mean Absolute Percentage Error (MAPE) metrics. Experimental results show that the LSTM model achieved superior performance compared to ARIMA, with RMSE and MAE values of 3.14 and 2.71, respectively, versus 6.05 and 5.21 for ARIMA. Product-level evaluation using MAPE indicates that Icy Lychee achieved a relatively low error rate of 37%, while Icy Mango (50%) and Icy Watermelon (52%) exhibited higher error rates, suggesting model performance may vary across product categories. These results demonstrate the LSTM model’s superior ability to capture nonlinear sales patterns compared to traditional statistical approaches. The model was integrated into a Django-based web system with a real-time visualization dashboard and sales logging features. The proposed system offers a practical and intelligent decision-support tool for retail inventory management, reducing stockout and overstock risks through data-driven forecasting.]]>
