<![CDATA[Dunio Petshop has difficulty estimating sales quantities, causing inventory decisions to rely mainly on judgment and physical stock checks. This study aims to develop a web-based goods sales forecasting system using the Prophet method. The dataset consists of historical sales of 20 pet-food products from January to December 2025. Development follows the CRISP-DM stages and integrates PHP, MySQL, and Python. The system produces 120 forecasts for January-June 2026. Black Box Testing confirms that all major functions operate as required. MAPE evaluation places 15 products (75%) in the Fair category and 5 products (25%) in the Poor category; Catlife Salmon records the lowest MAPE at 22.72%, while Smartheart Cat records the highest at 72.26%. The system can support stock-planning decisions, while a longer historical series is required to improve forecasting accuracy]]>
Copyrights © 2026
