<![CDATA[PT Atmadjaya Sembada Anugerah is a frozen food manufacturing company that faces challenges in stock management due to unpredictable daily fluctuations in product demand. Inaccurate demand forecasting can lead to inefficiencies in distribution and storage operations. This study aims to apply the Extreme Gradient Boosting (XGBoost) algorithm to forecast product demand using historical daily sales data. The process involves exploratory data analysis, data cleaning, feature engineering for time and statistical variables, and time-based data splitting. The model is trained using features selected through Recursive Feature Elimination and optimized using hyperparameter tuning with Optuna. Evaluation is conducted through TimeSeriesSplit cross-validation and assessed using three standard performance metrics. The results indicate that the model effectively captures seasonal patterns and general demand trends, although it remains limited in responding to sudden demand spikes. These findings support the use of XGBoost as a foundational approach for demand forecasting systems in stock planning, with potential for further improvement through the integration of external data and expanded feature sets.]]>
Copyrights © 2026
