Implementation of the Crisp-Dm Methodology and Naive Bayes Algorithm on A Raw Material Requirement Prediction System to Reduce Food Waste (Case Study: Adamsafee Bakery, Resto, & Cafe)

Accurate forecasting of raw material requirements is critical for culinary businesses to reduce food waste and optimize costs. In the case of Adamsafee Bakery, Resto, & Cafe, high levels of waste have been caused by reliance on intuition-based forecasting, resulting in both overstocking and understocking. This study develops a web-based predictive system using the Cross-Industry Standard Process for Data Mining (CRISP-DM) methodology and the Naive Bayes algorithm to classify demand patterns into three categories: high, medium, and low. Historical sales data were transformed into categorical attributes and processed through the Naive Bayes model to generate demand predictions. The system was evaluated by comparing predicted sales with actual outcomes. Results show that the model achieved an accuracy of 98.7% and a mean absolute percentage error (MAPE) of 1.31%, indicating that the forecasts closely aligned with real sales performance. These findings demonstrate the effectiveness of the Naive Bayes algorithm in supporting data-driven decision-making for inventory management. This data-driven approach replaces subjective decision-making, enabling management to optimize inventory, minimize food waste, and enhance operational efficiency and business sustainability, while also offering a baseline for future research using alternative machine learning algorithms.